(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "Two early Chinese bronze weapons with meteoritic iron blades"

TW;d; EARLY 

WITH 
METlORitiG IRONvBE 



R tl tH E RF O RD J, G ETT ENS 
ROY S.CLARKE, JR, 
W/T. CHASE 




OCGASjfpNAL PAi^ERS 
VOL. 4, NO. 1 



1 9 7 I 



FREE!?. GALiERY OF A1|T 
WASMpf<5T0N,^pX, ' ' 



TWO EARLY CHINESE BRONZE WEAPONS 
WITH METEORITIC IRON BLADES 



Two Chinese early Chou bronze weapons with iron blades 
Upper: ko 34.11, lower: ch'i 34.10. Slightly reduced. 



TWO EARLY 
CHINESE BRONZE WEAPONS 

WITH 

METEORITIC IRON BLADES 



RUTHERFORD J. GETTENS 
ROY S. CLARKE, JR. 
W. T. CHASE 




OCCASIONAL PAPERS FREER GALLERY OF ART 

VOL. 4, NO. 1 WASHINGTON, D.C. 

1971 



/G3 



Printed in Germany by 
BRODER HARTMANN 
BERLIN 



CONTENTS 



Page 



Foreword vii 

Preface ix 

List of illustrations xi 

List OF TABLES xii 

L Introduction i 

The weapons; their dates; their functions. 

IL Thebroad AXE 34.10 3 

Description; casting; sampling and analysis; metallographic 
structure of the bronze; corrosion products; the iron blade; inlay, 
radiographic examination; construction; repairs. 

IIL Thedagger AXE (^o) 34.11 15 

Description; chemical composition of the bronze blade; metal- 
lographic structure; corrosion products; the iron point; radio- 
graphic evidence of the method of construction; original appear- 
ance. 

IV. Studies on the oxide residues of the iron blade 

AND IRON POINT 23 



Introduction; remnant oxides of the blade of the broad axe 
34.10; remnant oxides of the point of the dagger axe 34.11; 
sampling; semiquantitative spectrographic analysis; chemical 
analysis; oxide composition; sampling for optical and electron 
microprobe examination; electron microprobe studies; octahe- 
drite structure and remnant oxide; remnant metal in the Wolf 
Creek meteoritic oxide; native iron; meteoritic origin of remnant 
oxides. 

V. Significance OF THE FINDINGS 57 

Introduction; statement by Ch'u Te-i; other weapons in the 
group; analyses of the weapons; comparison with other Hsin- 
ts'un weapons; date probably ca. 1000 B.C.; unpublished weapon 
34.14; Marquis K'ang; King Ch'eng's funeral; history of iron in 
China; other meteorite falls; other uses of meteoritic iron; the 
casting-on technique. 

Bibliography 73 



V 



FOREWORD 



The presence of iron blades in these two early Chou bronze 
weapons intrigued the late Archibald Gibson Wenley from the 
time he began to study them while working on the first catalogue 
of Freer Chinese bronzes that came out in 1946. In 1950 he asked 
the late Dr. William F. Foshag of the Department of Geology, 
Smithsonian Institution, to examine them from the scientific point 
of view; and the results of that investigation gave further stimulus 
to the founding of the Freer Technical Laboratory which was then 
in the planning stage and which was to open the following year 
under the direction of Rutherford John Gettens. 

In spite of the calls for help that flooded the new laboratory 
from every side, Mr. Gettens returned to these two weapons from 
time to time; and during the ten years when the study of the cere- 
monial vessels was in progress much additional related informa- 
tion came to light. Roy S. Clarke, Jr. of the Division of Mete- 
orites, Smithsonian Institution, and W. T. Chase, Assistant Cu- 
rator of our laboratory and now Mr. Gettens's successor as its 
director, made important contributions at every stage of the 
work; and in recognition of this the book is published as the col- 
laborative work of the three authors. 

We are pleased at long last to publish this small volume, and 
it is only fair to point out that its size is in no way related to the 
immense amount of work that went into it. It should shed new 
light on the knowledge of metal technology in ancient China and 
also on other phases of the cultural history of that country some 
3,000 years ago. 

Freer Gallery of Art John Alexander Pope 

Washington, D. C. Director 
October, 1970 



Vll 



PREFACE 



The authors are grateful to a number of persons who have 
helped in this study. First to Daniel Polansky and associates at 
the U.S. Naval Ordnance Laboratory, White Oak, Maryland for 
the radiographs which gave major impetus to the study in its early 
stages. Mrs. Elisabeth West FitzHugh made some of the bronze 
analyses and did much of the early X-ray diffraction work. Mrs. 
Ilona V. Bene also contributed to the chemical analyses. Dr. 
Michael B. Duke of the U.S. Geological Survey made the early 
electron microprobe exploratory studies. Joseph Nelen of the Di- 
vision of Meteorites of the National Museum of Natural History 
obtained all of the electron microprobe data reported here. Dr. 
Kurt Fredriksson, also of the Division of Meteorites, gave valu- 
able advice in connection with the probe work. Professor Paul 
Ramdohr of the University of Heidelberg, a specialist in ore micro- 
scopy, studied the polished sections of the two weapons. Dr. Jo- 
seph I. Goldstein of Lehigh University also participated in the 
discussions and contributed ideas. 

Several have generously read the manuscript and made con- 
structive comment: Dr.V. F. Buchwald of the Technical Univer- 
sity, Copenhagen; Dr. Brian Mason of the Department of Min- 
eral Sciences and Robert M. Organ of the Conservation- Analyti- 
cal Laboratory, Smithsonian Institution; Dr. Cyril S. Smith of 
the Massachusetts Institute of Technology; and Dr. Thomas Law- 
ton of the Freer Gallery. Grover Moreland, also of the Depart- 
ment of Mineral Sciences, prepared the metallographic mounts 
used in the study of the dagger axe. We are grateful to Raymond 
Schwartz of the Freer Gallery for much of the fine photographic 
work, and to Harold Westley of the Conservation-Analytical 
Laboratory, U.S. National Museum, for most of the spectro- 
graphic data presented here. The fact that this book has come out 
at all is in no small measure due to the patience and skill of our 
Editor, Lloyd E. Langford, who supervised every detail of pre- 
paring it for the printer and seeing it through the press. 



IX 



The study was partially supported by NASA Grant NsG- 
688. 



Rutherford J. Gettens 
Roy S. Clarke, Jr. 
W. T. Chase 



X 



LIST OF ILLUSTRATIONS 



Frontispiece. Color plate of the two weapons (c^'i, FGA 34.10 
and ko, FGA 34.1 1). 



Figure Page 

Ch'i 34.10 

1 . Two overall views 4 

2. End view of blade socket 6 

3. Side view showing mold flash 6 

4. Metallograph of naturally etched surface 7 

5. Radiograph 10 

6. Reconstruction 12 

7^034.11 

7. Two overall views 14 

8. Metallograph of naturally etched surface 16 

9. Residues of wooden handle 16 

I o. End view of remains of point 17 

11. Radiograph 18 

12. Reconstruction 20 

13. Laminated structure in iron rust 22 

14. A drill hole filled with ferric chloride 22 

15. Metallographic mount of specimens from 34.10 32 

16. Metallographic mount of specimens from 34.1 1 33 

17. Metallographic mount of specimens from 34.11 34 

18. Detail of figure I j 37 

19. Detail of figure i ^ 38 

20. Detail of figure 16 39 

21. Microprobe trace from /ig/^ re 7 5 42 

22. Microprobe trace from figure 79 44 



xi 



23. Microprobe trace from figure 20 45 

24. Widmanstatten pattern in the Arispe Meteorite 47 

25. Phase diagram 49 

26. Photomicrograph from Campo Dei Cielo Meteorite .... 50 

27. Microprobe trace from figure 26 52 

28. The set of twelve weapons 59 

29a and b. Two views of weapon 34.14 64 



LIST OF TABLES 



Table 1 . Semiquantitative spectrographic analyses of iron 

oxide samples 26 

Table 2. Chemical analyses of oxide samples 28 

Table 3. Analyses of magnetic and nonmagnetic fractions 

of broad axe (34.10) blade oxide 28 

Table 4. Electron microprobe analyses of metallic inclu- 
sions in oxides 41 

Table 5 a. Wet chemical analyses of the set of bronze 

weapons 61 

Table 5 b. Spectrographic analyses of the set of bronze 

weapons 61 



xn 



1. INTRODUCTION 



The Freer Gallery of Art owns two unique metal weapons 
from ancient China (frontispiece). Though they have rarely been 
exhibited these weapons are shown in the old Freer catalogue 
(Freer Gallery of Art, 1946). One (34.10) is a broad axe (ch'i) 
with a bronze tang (nei) and the rusted remains of an iron blade; 
the other (34.1 1) is a dagger axe (ko) with a bronze blade and the 
remains of an iron point. Both were acquired in 1934 during the 
directorship of John Ellerton Lodge as part of a group of 12 
bronze weapons said to have been found in 193 1 in Honan prov- 
ince in China and to date from the early Chou dynasty (about 
1 000 B.C.). These two weapons are of high interest to the historian 
of technology for three reasons: (i) the combination of two dif- 
ferent metals, one common at the time of manufacture and the 
other presumably unknown in China; (2) evidence that the two 
metals were joined by the casting-on method, a well-known tech- 
nique of the Bronze Age in China; and (3) evidence that the iron 
in both weapons is of meteoritic origin. 

Much of the information which points to the above conclu- 
sions comes from the use of a variety of instrumental techniques 
which in recent years have been widely applied to problems of 
art and archaeology: X-radiography, X-ray diffraction, electron- 
beam micro-spectrometry and emission spectrometry were all 
employed. The evidence from each technique will be introduced 
in its proper place. First, however, we must consider briefly the 
provenance, date and use of these two objects and describe them; 
a more complete discussion of the art historical questions these 
weapons raise and their place in the history of Chinese technology 
appears below in Section V. 

Both weapons are well-known types and they can now be 
dated to about 1000 B.C. Magdalene von Dewall (1967, p. 527) 
dates them in the tenth century B.C. — very soon after the Chou 
conquest. Max Loehr (1956, p. 209) illustrates and describes sev- 
eral Shang or early Chou all-bronze broad axes of similar style 



I 



and shape to our ch'i (yiieh), and also several weapons in the style 
of our ko. Most are solid bronze but a few have jade blades; none 
of early date shown by Loehr have iron blades or points except a 
pair of bronze axes provisionally dated 770-450 B.C. with re- 
mains of iron pikes that look like handles. The kind of iron is not 
mentioned. There is scattered mention, however, of contempo- 
rary Luristan and Scythian swords with iron blades. 

It might be presumed that the iron blades were attached to 
the weapons to make them more functional, but because of the 
scarcity of iron in the early Chou period, and the decor on the 
weapons, especially the inlay on the ch'i, it seems they were prob- 
ably made for ceremonial purposes. 



2 



11. THE BROAD AXE (C//7) 34.10 



The object (fig. /), which is 17.1 cm. long and 10.8 cm. wide, 
weighs 437.5 grams. The upper part of the blade and the tang 
(nei) is of a single piece of bronze; the blade or cutting edge is 
iron which was originally fashioned so that it could be secured 
in the edge of the bronze. The iron blade is apparently completely 
converted to rust (fig. 2) and has broken away from the tang. In 
the center part of the bronze axe head is a hole 1.8 cm. in diam- 
eter; and around this is cast, in narrow sunken lines, a stylized 
dragon form with eyes in low relief. In the end of the tang is also 
cast a fao fieh mask in relief with deep channels which appar- 
ently were originally inlaid with tesserae of reddish-brown mate- 
rial of which only tiny fragments still remain (see below). Two 
lashing slots in the back probably received thongs for hafting and 
a small "splint hole" in the tang may have held a transverse pin 
to help fasten the lashing. The bronze surface is mostly covered 
with blue and green copper corrosion products but in some areas, 
especially that part which was protected by the haft, the original 
polished metal surface still shows. The iron blade appears to be a 
complete mass of iron rust. 

The bronze tang is cast in one piece, probably in a two-piece 
mold. Mold marks show distinctly along the upper edges of the 
broad head and at the corners where the tang joins the head 
{fig. j). There are no mold marks on the inside of the smaller hole 
or on the inside of the two slots. These appear to have been drilled 
or cut in the bronze but it is also possible that they were cast. The 
sides of the large hole are concave. The long recess or slot to fit 
the iron blade appears to be cast in; the slot is still partially filled 
with iron rust and in the radiograph (fig. ^, see below) the irregu- 
lar mottled area shown between edge and hole indicates the slot 
is about 1.5 cm. deep. 

Samples for analysis were taken by hand drilling into the edge 
of the tang with a No. 44 (.086 inch) steel twist drill. After sam- 
pling the drill hole was plugged with threaded copper wire and 



3 



Fig. 1, — 34.10. Two views of a Chinese bronze broad axe with iron blade of the type ch'i formerly inlaid; dated 
early Chou dynasty which began in 1027 B.C. It is said to have been dug up with other ancient weapons by a 
native at Hsiin Hsien, Wei-hui-fu, Honan province, not far from An-yang district. Length, 17. i cm. 



the location of the spot concealed. Wet analysis for major con- 
stituents shows: 



Cu 81.8 % (electrolytic) 

Sn I 5.8 % (gravimetric) 

Pb trace 

Total 97.6 % 

The analysis is the average of duplicate determinations made on 
samples weighing approximately 80 milligrams each (0.08 grams; 
see also Table 5 a). 

Spectrographic analysis to estimate minor and trace constitu- 
ents was carried out on a Jarrell-Ash 3 m. grating spectrograph. 
The analysis made on duplicate lomg. samples of bronze taken 
from the core shows, in addition to copper and tin, the following 
elements: 

In the range 0.0 i-o. i % Bi, Fe, Ni, Sb, Si, Pb 

In the range o.ooi-o.oi % Ag, Co 
Less than 0.001% Cr, Mg 

Sought but not found Al, As, Mn, Zn 



Metallographic structure of the bronze 

No attempt was made to take a section of the bronze for 
metallographic polishing and etching. Scattered areas of the sur- 
face are not encrusted with corrosion products but are still me- 
tallic in appearance. Examination at X30 magnification shows, 
however, that corrosion has begun in these areas sufficiently to 
etch the metal and to reveal its crystalline structure. The darker 
dendritic forms of high-copper low-tin alpha bronze stand out 
against the lighter areas of tin-rich alpha which have been par- 
tially corroded, probably to hydrous tin oxide, H^SnO^ {fig. 4). 
This is clear evidence within the metal itself that it was formed 
by casting. 



5 




Fig. 2. — 34.10. Chinese bronze broad axe of the type ch'i with iron blade. End view (x 2) showing laminated 
structure of the iron rust retained inside the bronze ferrule. 




Fig. 3. — 34.10. Chinese bronze broad axe of the type ch'i with iron blade. 
Detail (x 1.7) showing a ridge along edge of the tang which is the flash 
(mold mark) left at the juncture of the two halves of the mold in which 

the axe was cast. 



Corrosion products 



Much of the bronze surface is thinly encrusted with malachite, 
Cu2(OH)2C03, and azurite, Cu3(OH)2(C03)2. The edge of the 
break across the middle reveals that in places the metal is pene- 
trated clear through with cuprite, CugO. 

Some fragments of wood are still attached to the edge of the 
iron (fig. /). The wood could not be identified because it is so 
rotted that no diagnostic features remain; not even enough to 
identify the class of wood. 




Fig. 4. — 34.10. Chinese bronze broad axe of the type cb'i with iron blade. Met- 
allograph (X75) of the naturally etched surface without preparation in any way. 



The iron blade 

When the axe first came into the collection the iron blade and 
the bronze tang were apparently still joined. They eventually be- 
came separated and traces of animal glue at the break revealed 
an old repair. The stump of the blade that still remains in the slot 
gives convincing evidence that the blade and head belong together. 



7 



Although the blade is now a mass of iron rust much scaled and 
blistered on the surface, it appears to retain quite well its original 
shape. The fractured edges of both blade and stump show a lami- 
nated structure (fig. 2). In 1950 the late Dr. William F. Foshag 
of the Department of Geology, Smithsonian Institution, examined 
the iron residues and reported, "Sample 34.10 was strongly mag- 
netic containing considerable nickel. This resembles in many re- 
spects some oxidized meteoritic iron." (See FGA folder sheet for 
this object.) This remark about meteoritic iron was the clue that 
started the long investigation which will be described later in this 
report. 

The inlay 

The high relief decor on each side of the axe head and the 
fao fieh surrounding the central hole was originally inlaid with 
some sort of material of which only traces remain. Small, reddish 
brown chunks are visible in the high relief design; one can be seen 
in figure i (lower), just below and to the left of the left eye of 
the mask on the haft. The green inlay in the lower design can 
be clearly seen in the frontispiece. When the red chunks in the 
haft are viewed under a low-power stereomicroscope (at magnifi- 
cations about X30) they are seen to have a conchoidal fracture, 
and the fractured edges are green. X-ray powder diffraction of a 
number of samples from this area shows some lines which could 
be attributed to tridymite along with other lines which we cannot 
identify. All of the patterns are weak and diffuse and show much 
fluorescence. Powder patterns were also taken from some of the 
inlay on the lower portion of the weapon, and these are very 
similar to those from the upper part. All of the inlay material is 
of similar crystal structure, and it is definitely not turquoise or 
nephrite. 

A sample from the red inlay was removed with a vibrating 
steel needle, mounted in Canada balsam, and examined micro- 
scopically. It consists of fractured particles with brownish inclu- 
sions which are so small that they are difficult to resolve with the 



8 



optical microscope; they appear to be polygonal grains. The matrix 
material is isotropic and has a refractive index so close to that of 
balsam (1.53) that its edges are invisible. The inclusions have a 
higher refractive index than the matrix and are birefringent. This 
evidence tends to make one conclude that this material is a glass 
with crystalline inclusions. 

A qualitative spectrographic analysis of the material supports 
this conclusion. Two samples were taken from the reddish ma- 
terial, and both contained Si, Fe, P, Mg, Al, Ca, and Cu. No Pb 
was detected. Fe and Cu might easily have come from the close 
proximity with these materials in the weapon. If it is a glass, it is 
a lime-silica glass and not a lead glass, a fact which the radiograph 
confirms. 

If one accepts the inlay in this weapon as being glass, and there 
seems to be no reason to doubt it, the weapon is rendered even 
more remarkable. Glass beads have been discovered in ninth- 
eighth century B.C. tombs in China (Cheng, 1963, p. 198), so this 
object is not a unique use of glass at this period. The inlay in other 
Shang and Chou weapons should be critically examined to be sure 
that those labelled "turquoise" are not actually glass. 



Radiographic examination 

The whole broad axe was X-rayed at the U.S. Naval Ord- 
nance Laboratory near Washington, D.C. The radiograph (see 
caption, fig. j) reveals various details of structure and condition: 
(i) There appears to be no large trace of metallic iron in the mass 
of rust which is all that remains of the exposed part of the iron 
blade. (2) There is no distinct terminal line for the iron blade, only 
an indistinct mottled area just inside the slot. (3) There are three 
nearly evenly-spaced spots of greater density on the edge of the 
bronze which seem to relate to the device that was used to secure 
the iron blade to the tang. (4) The bronze tang has been broken 
across the middle and repaired at the edges with tin-lead solder 
(see above). 



9 



Fig. 5. — 34.10. Radiograph of bronze broad axe. The view reveals that the bronze head had previously been 
broken into three pieces and joined with solder. It also shows three spots of high density along the ferrule edge 
which were later revealed to be, not rivets, but cast-in cross bars made to secure the iron blade to the bronze 
head. They indicate also that the bronze head was "cast-on" to the iron. Since the iron blade is almost completely 
converted to oxide, it hardly shows in the original radiograph, and its appearance has been enhanced photo- 
graphically here. Radiograph taken at 150 Kv, 10 Ma, distance from target 76 cm.; developed in Eastman Kodak 
Microdol developer. Courtesy, U.S. Naval Ordnance Laboratory, White Oak, Silver Spring, Maryland. 



Construction 



A special study was made to determine how the iron and 
bronze members were locked together. It was previously men- 
tioned that the radiograph reveals three nearly evenly-spaced 
white spots in the edge of the slot which suggest that rivets of some 
dense metal were used to secure the iron blade to the head. Probing 
of the area of one of the spots on the inside of the slot revealed 
there are two stumps or spurs of bronze which are opposite each 
other, and moreover they seem to be part of the bronze casting 
itself. There is no evidence of them on the bronze surface at the 
edge of the slot. Apparently they are not rivets. The location of 
the other white spots in the radiograph could not be probed, be- 
cause of the hardness of the iron rust covering them. The evidence 
here indicates that the bronze tang was cast on to an iron bit that 
had already been wrought from iron. It further indicates that 
prior to the joining, the iron blade was drilled through along the 
joining edge with three holes. The mold for the bronze tang was 
then built around the wide end of the bit and the tang was cast on. 
The molten bronze ran into the holes in the iron and when solidi- 
fied, the blade and tang were securely locked together (/ig. 6). The 
interlock casting of bronze legs and handles to bronze vessels, or 
vice versa, was well known in China in the latter part of the Chou 
dynasty and here it seems already to be a developed technique of 
joining in early Chou (Gettens, 1969, ch. 4). 



Repairs 

The radiograph showed a break and lead solder repair on both 
sides of the bronze tang, continuous with the upper line of the 
cut-out decoration {fig. j). Later the repair join was taken apart 
and this revealed that the broken edges had been filed, perhaps to 
make them join better and to permit soldering. Much cuprous 
oxide along the fracture line also indicates that this is an old break 
which occurred perhaps at the time of excavation. The metal was 
too corroded to be effectively joined by tin-lead solder. The pres- 



1 1 




Fig. 6. — 34.10. Reconstruction of the ceremonial broad axe with iron blade shows how the 
bronze tang was cast on and locked to the iron blade. 



ence of the repair was outwardly concealed with daubed-on paint 
in which the modern artificial pigments, Prussian blue and emer- 
ald green (Paris green), were identified. The concealment of the 
repair was aided by scattered splashes of earthy material applied 
to the surface. These repairs were perhaps made in China before 
the axe was placed on the antique market. 



13 




Fig. 7. — 34.11. Two views of a Chinese bronze dagger axe of the type ko originally with iron point; dated early 
Chou dynasty which began in 1027 B.C. It is said to have been dug up with other ancient weapons by a native 
at Hsun Hsien, Wei-hui-fu, Honan province, not far from An-yang district. Length, 18.3 cm. 



III. THE DAGGER AXE (KO) 34.11 



The object {fig- 7), which is 18.3 cm. long and 7.0 cm. wide, 
weighs 378.5 grams. The blade and tang are cast in bronze, but 
the point which apparently was originally made of iron, is rusted 
away and mostly lost. The blade is decorated on either side with 
a bold dragon motif cast in low relief and the end of the tang is 
also twice decorated with similar stylized dragon motifs. The tang 
end of the blade begins in a crosswise hafting ridge, and the inner 
part, which was covered formerly by a wooden shaft or handle, 
is flat and undecorated. A plain round hole and a rectangular slot 
in the heel of the blade probably served to lash the blade to the 
haft with thongs. The bronze surface is mostly covered with red, 
blue and green copper corrosion products which are quite deco- 
rative. The broken-off end of the blade {fig. 10) is encrusted with 
iron rust, evidence that the blade originally had an iron point. 

Chemical composition of the bronze blade 

Samples of the bronze were taken by drilling into the edge of 
the tang with a No. 44 steel twist drill; after drilling, the hole was 
plugged with threaded copper wire and the spot concealed. Wet 
analysis for major constituents shows: 

Cu 85.5% (electrolytic) 

Sn 12.2 % (gravimetric) 

Pb 2.1 % (electrolytic) 

Total 99.8% 

The analysis is the average of duplicate determinations made 
on samples weighing approximately 80 milligrams each (see also 
Table 5). 

Spectrographic analysis on duplicate lomg. samples of bronze 
taken from the core shows in addition to copper, tin and lead the 
following elements: 



15 




Fig. 9. — 34.11. Chinese bronze dagger axe. Residues of the wooden handle which was lashed to the bronze axe 
head still adhere to the copper corrosion products on the surface. 



In the range o. i-i .o % Ag, Al, As 

In the range o.o i-o. i % Co, Fe, Ni, Sb 

In the range o.ooi-o.oi % Bi, Si 

Less than <o.ooi % Cr, Mg 

Sought but not found Mn, Zn 

The bronze alloy is definitely different from the alloy of the 
broad axe, but both are made of moderately high tin alpha bronze. 
Lead content is low which is in contrast to the highly leaded 
bronze usually found in the ceremonial vessels of this period. 



Fig. 10. — 34.11. End view (x 2) of Chinese bronze 
dagger axe, showing the iron rust remains of the 
original iron point. 




17 



Fig. 11. — 34.11. Radiograph of the bronze dagger axe. The radiograph 
shows that the iron rust deposit at the point end conceals an open 
dragon's mouth and that the bronze blade was cast-on to the point 
which had been shaped in the form of a key so that bronze and iron are 
securely locked together. Radiograph taken at 250 Kv, 10 Ma, distance 
from target 183 cm.; developed in Eastman Kodak Microdol developer. 
Courtesy, U.S. Naval Ordnance Laboratory. 



Metallographic structure 

A small unencrusted area on the tang indicates that the metal 
was originally well polished to an almost mirror-like surface but 
ghosts of dendritic structure show in the little-corroded areas of 
the surface (fig. 8) which further indicates the bronze was cast. 

Corrosion products 

The bronze surface is covered in certain areas with patches 
of blue azurite and green malachite. A fairly solid layer of cu- 
prite underlies the basic copper salts. No atacamite (basic copper 
chloride, Cu2(OH)3Cl) was found. The high total percentage of 
Cu, Sn and Pb in the chemical analysis indicates that corrosion has 
not penetrated deeply into the metal core. Neither tin nor lead 
corrosion products show on the surface. The undecorated area on 
one side of the tang is traversed with parallel striations crosswise 
on the blade which were recognized as impressions or residues of 
the original wooden haft (fig. 9). There were not sufficient diag- 
nostic features left in the wood cell structure to identify the species 
or even the genus of the wood. 

The iron point 

As previously stated no actual point exists; only the rusty 
residues of what is presumed to have been a point at the end of 
the bronze blade. Dr. W. F. Foshag examined, in 1950, samples of 
these rusty accretions at the same time he examined the rusty blade 
of the broad axe and found that, "Sample 34.1 1 was not magnetic, 
contained some copper; but no nickel was detected. This specimen 
of oxidized iron is not believed to be from a meteorite." (SeeFGA 
folder sheet for this object.) The specific studies on the iron resi- 
dues, which will be reported later, indicate strongly, however, 
that the original iron of this weapon was also meteoritic in origin 
and they explain why Dr. Foshag failed to reach that conclusion. 



19 





Fig. 12. — 34.11. A reconstruction of the Chinese bronze dagger axe, shows how the iron 
point looked originally and how the bronze blade was cast-on and locked to the iron. 



Radiographic evidence of the method of construction 



This axe was also X-rayed at the U.S. Naval Ordnance Labo- 
ratory and the radiograph revealed some striking features of con- 
struction {fig. 1 1). It shows that the bronze blade terminates in the 
gaping mouth of a dragon which grasps the iron point in its teeth. 
Less dense areas just back of this open mouth show that the butt 
of the iron point terminated in a double notched tang or key by 
which it was secured to the blade. Small drill holes made in this 
area confirmed the presence of a layer of iron oxide sandwiched 
between the two outer layers of bronze. It is probable that the 
mold for the bronze blade was built around the keyed end of the 
iron point so that when molten bronze flowed in and around the 
key and solidified, the blade and point were securely locked to- 
gether. Again we have an example of lock-on casting. In the re- 
construction of the axe {fig. 12) it appears that the dragon is 
swallowing the iron point. 

The bronze blade is cast in one piece probably in a bivalve 
mold. The only signs of mold marks are on the square ends of the 
hafting ridge. There are no mold marks on the inside of the thong 
holes. 



Original appearance 

It is difficult to say just how the dagger-axe looked originally, 
but the iron point was probably about 7-8 cm. long, making the 
total length of the weapon about 23-25 cm. {fig. 12). There are in 
existence many all-bronze dagger axes of this same style and peri- 
od. From pictures on tomb tiles of the Han period, and from other 
sources we can imagine that the handle of the axe was a simple 
staff with a slot cut in near the end to accommodate the tang and 
that the dagger axe was secured to the staff by thongs laced 
through the slot and hole. Max Loehr (1956, ch. Ill) shows illus- 
trations of Shang pictographs which give some idea of the way 
dagger axes were hafted. 



21 



Fig. 13. — 34.11. Chinese bronze dagger axe. The macrograph (x lo) shows traces 
of a laminated structure in the iron rust which is characteristic of the mineral 
goethite. This small area of laminated structure occurs about 2 cm. from the end 
of the bronze blade in an area of rust that was transported and deposited on the 

bronze surface. 




Fig. 14. — 34.11. Chinese bronze dagger axe. The macrograph (x 9.2) shows a drill hole that was made near the 
end of the bronze blade passing through the outer encrustation of iron rust, through the bronze layer and into 
the inner iron core. After about five years in storage, a crusty brownish globular shaped deposit of ferric chlor- 
ide formed in the drill hole. 



IV. STUDIES ON THE OXIDE RESIDUES 
OF THE IRON BLADE AND IRON POINT 



Early in the study it became evident that the iron residues are 
of special interest, not only because the weapons date from a peri- 
od which precedes the accepted date for the beginning of the iron 
age in China, but also because Dr. Foshag made the inference in 
his early findings that the iron of the broad axe contains enough 
nickel to suggest that it might be of meteoritic origin. This interest 
initiated studies which, over a period of some years, have con- 
firmed the meteoritic origin of the iron of the broad axe and have 
also indicated that the iron of the dagger axe probably had a simi- 
lar origin. 

Remnant oxides of the blade of the broad axe 34.10 

The iron blade which served as the cutting edge has been dras- 
tically altered by corrosion. The radiograph which reveals many 
details of the bronze tang is very dark opposite the area of the 
blade indicating almost complete absence of metallic iron. X-ray 
powder diffraction analysis of samples of rust showed that it con- 
sists of two hydrated iron oxides, goethite, alpha FeO(OH) and 
lepidocrocite, gamma FeO(OH), dimorphous with goethite, and 
three anhydrous oxides, magnetite, Fe304, maghemite, gamma 
Fe203, and hematite, alpha Fe203. The rusty mass is sufficiently 
magnetic to permit the blade to be raised free from the table with 
a strong magnet. 

Remnant oxides of the point of the dagger axe j4.11 

The rust residues cover the blunt end of the bronze blade in an 
irregular crusty mass. X-ray diffraction analysis of the outer ex- 
posed crusts shows presence of only the hydrated oxides goethite 
and lepidocrocite. (This is apparently where Dr. Foshag got his 
sample.) One small area (fig. i j) has ripple marks which are 
characteristic of naturally formed goethite. There is a cleft in the 



23 



end of the blade which divides the iron rust in two parts {fig. lo). 
There is no obvious reason for this but it may demonstrate the 
completeness of the transfer of iron salts from their initial posi- 
tion, or it may simply be geometric distortion caused by the oxida- 
tion of the iron with expansion of volume. The rust-covered end 
of the blade rises to a strong magnet. Since goethite is not magnetic 
it was suspected that the remnant oxides in the interior in the 
neighborhood of the key revealed by the radiograph might be 
magnetite. To test this idea small exploratory borings were made 
into the interior, which confirmed the presence of iron oxide; and 
furthermore. X-ray diffraction analysis showed that the iron ox- 
ide deep in its interior is magnetite, not goethite. This explains the 
attraction to the magnet of the end of the bronze blade. Appar- 
ently the conditions in the interior of the blade favored formation 
of magnetite over goethite. 

The exploratory drill holes were made in 1956 before the in- 
stallation of air-conditioning in the gallery. After about five years 
in storage it was observed that a bubble-like reddish brown crust 
had formed in the drill holes (fig. 14). The crust was glossy and 
amorphous, like a solidified gel; it dissolved in dilute nitric acid 
and tested strongly for chloride ion. Apparently unstable iron 
chlorides are present and since these are strongly hygroscopic they 
may account for the transport of iron compounds from the iron 
point to the bronze surface. The presence of iron chloride may also 
explain the rapid disappearance of most of the original iron point. 
It is strange, however, that no evidence of chloride salts were ob- 
served among the copper corrosion products. 

Sampling 

At this point it was decided that more precise information was 
needed about the composition of the iron oxides of both weapons. 
Small chips were removed from the broken-off end of the iron 
blade and borings were made into the split end of the dagger axe 
in an attempt to get oxide from deeper in the interior where it 
should be less altered. These samples were ground to a fine powder 



24 



in an agate mortar and subsamples were taken for X-ray diffrac- 
tion analysis (see above) and spectrographic and chemical anal- 
ysis. The total amounts of sample prepared for analysis were 580 
mg. for the broad axe (34.10) and 740 mg. for the dagger axe 
(34.1 1). (It will be more convenient from now on to report labo- 
ratory findings on the iron oxides of both weapons together than 
to keep them separate as has been done previously.) 

A subsample of each oxide sample was examined for strongly 
magnetic material using a hand magnet. The broad axe blade 
(34.10) sample yielded 72 percent by weight of magnetic material 
and a non-magnetic fraction. X-ray diffraction analysis indicated 
that the magnetic material was largely magnetite with small 
amounts of maghemite, goethite and hematite. Only goethite and 
lepidocrocite were indicated for the nonmagnetic fraction. Mag- 
netic separation of the dagger axe sample indicated less than one 
percent magnetic material, too little to examine further. 

Semiquantitative spectrographic analysis 

Approximately 30 mg. portions of both oxide samples were 
submitted to the U.S. Geological Survey for semiquantitative 
spectrographic analysis. The elements that were detected and their 
approximate concentration ranges are listed in Table i. The 
spectrographic procedure used has been described by Waring and 
Annell (1953), and their paper lists detectability data for indi- 
vidual elements. 

Chemical analysis 

The oxide samples described above were quantitatively ana- 
lysed for their major constituents and the results are given in 
Table 2. Water was determined by the Penfield method without 
flux on separate 158 mg. (broad axe, 34.10) and i5omg. (dagger 
axe, 34.11) samples. The other constituents were all determined 
on single samples of 228 mg. (34.10) and 205 mg. (34.1 1). 

The oxide samples were dissolved in dilute HCl for major con- 
stituent analysis. A closed system swept with nitrogen was used 

^5 



TABLE I 

Semiquantitative spectrographic analyses 
of iron oxide samples* 





I 


II 


Element 


Broad axe 


Dagger axe 




(34.10) t 


(34.11) t 




percent 


percent 


Si 


0.7 


0.5 


Al 





.001 


Fe 


M 


M 


Mg 


.005 


.03 


Ca 


.015 


' .03 


Ti 


.0015 


.005 


Mn 


.0003 


.0005 


Ag 





.00007 ( 


Ba 


.05 


.0015 


Be 


.00007 


.0007 


Co 


.15 


.05 


Cr 


.0005 


.0007 


Cu 


0.3 


0.5 


Ge 


.005 


.005 


Ni 


3.0 


0.5 


Pb 


.007 


.003 


Sn 


.01 


0.15 


Sr 


.0007 






* Analyst: Nola Sheffey, U. S. Geological Survey. 

f Results are reported in percent to the nearest number in the series 
1, 0.7, 0.5, 0.3, 0.2, 0.15 and 0.1, etc: which represent approximate mid- 
points of group data on a geometric scale. The assigned group for 
semiquantitative results will include the quantitative value about 30% of 
the time. 

M = major constituent — greater than 10% 
O = looked for but not detected 

Note. The following elements were looked for but not detected in either 
sample: Na, K, P, As, Au, B, Bi, Cd, Ce, Ga, Hf, Hg, In, La, Li, Mo, 
Nb, Pd, Pt, Re, Sb, Se, Ta, Te, Th, Tl, U, V, W, Y, Yb, Zn, Zr. 



26 



and the evolved gas passed through a solution of lead acetate. No 
precipitate formed, indicating the absence of significant quantities 
of acid-soluble sulfide. The insoluble residue from the sample so- 
lution was a colorless material which was filtered, ignited and 
weighed. On ignition the residue from the broad axe (34.10) oxide 
developed a slight brown color, while the residue from the dagger 
axe (34.1 1) oxide became rose colored. X-ray powder analysis of 
these residues indicated that alpha quartz was the major constitu- 
ent. The sample solution was evaporated to dryness in glass after 
the addition of 5 ml. of HNO3. Acid soluble Si02 was dehydrated 
by this treatment, collected on paper, ignited, weighed, and the 
loss in weight determined after HF volatilization. The HCl solu- 
tion from the separation of SiO^ was taken to 100 ml. in a vol- 
umetric flask, and this solution was used as a working solution 
for the determination of other constituents. Ni was determined 
on a 50 ml. aliquot using a standard gravimetric method, precipi- 
tation with dimethylglyoxime from a tartaric acid solution. Total 
iron was determined on a 20 ml. aliquot by titration with standard 
KoCrgOy after reduction by a silver reductor. Co was determined 
on a 5 ml. aliquot using the nitroso-R method as described by 
Moss, Hey and Both well ( 1 96 1 ). 

Three additional small samples from the dagger axe (34.11) 
oxide were partially analysed by the same general procedures in 
an unsuccessful attempt to find material of higher Ni content. 
These results are also reported in Table 2. Approximate sample 
weights used were III, 185 mg; IV, 240 mg; and V, i3omg. A 
50 ml. aliquot of 100 ml. of solution was used for the Ni determi- 
nation. Sample IV was also separated with a magnet, and a 2 
percent strongly magnetic fraction was recovered, confirming a 
similar measurement mentioned above. X-ray diffraction analysis 
of this fraction indicated magnetite and maghemite with small 
quantities of goethite and hematite. The nonmagnetic fraction, the 
bulk of the sample, contained mainly goethite with some lepido- 
crocite and a little hematite. 

Analysis I in Table 2 is presented in two ways. In (a) all of 
the Fe is calculated as Fc^O;}. This leads to a high total and con- 



^7 



TABLE 2 





Chemical analyses 


of oxide 


samples* 






I 




TT 


TTT 


TV V 

IV V 




Broad 




F)?l 0"0"Pt* 






axe 




axe 


axe 


axe axe 




(34.10) 










(a) 


(b) 








Ignited insoluble 












residue 


2.6% 


2.6% 


1.2% 


0.5% 


0.3% 0.3% 


Acid soluble SiOa 


0.1 


0.1 


1.1 








74.1 




OZ,. J 






NiO 


8.53 




0.80 


1.1 


1.3 0.74 


(Ni, Fe) Fe^O^ 




80.8 








CoO 


0.16 


0.16 


0.05 






Total HoO 


16.0 


16.0 


13.6 






Totals 


101.49 


99.66 


99.1 






Cu(spectrographic) 


0.3 


0.3 


0.5 







* Analyst: Roy S. Clarke, Jr. 

f Chips from edge of fissure. 

4: Drillings inside fissure. 

§ Deeper drillings inside fissure. 



TABLE 3 

Analyses of magnetic and nonmagnetic fractions 
of broad axe (34.10) blade oxide* 

Nonmagnetic Magnetic Composite I (a) 

portion portion (from 

59 mg. 151 mg. Table 2) 



Ignited insoluble 


4.2% 


2.0% 


2.6% 


2.6% 


residue 










FeoOg 


66 


76 


74 


74.1 


Nib 


8.0 


8.6 


8.5 


8.53 


CoO 


.26 


.20 


.21 


.16 


Loss on ignition 


19 


13 


15 




Total HoO 








16.0 



* Analyst: Roy S. Clarke, Jr. 



28 



flicts with the X-ray data which indicates that most of the Fe is 
present as magnetite. In (b) all of the Fe and Ni are accounted 
for in the magnetite formula, (Ni, Fe) Fe204, and a more satisfac- 
tory total results. Calculation (b) is undoubtedly nearer to the 
correct presentation than (a). The true value, however, must be 
between these extremes. 

The magnetic (151. 3 mg.) and nonmagnetic (59.4 mg.) frac- 
tions of sample I (broad axe, 34.10) were also analyzed by the 
same general procedures used above. The results of this partial 
analysis are given in Table 3. Because of the small samples avail- 
able loss on ignition was determined instead of total HoO, simply 
by moderate heating over a Meker burner for approximately 1 5 
minutes. This determination gave information similar to the H^O 
dermination made previously and permitted the same sample to 
be used for all of the determinations reported. The residue from 
the ignition was dissolved with prolonged digestion in mixed HCl- 
HNO3 and finally evaporated to dryness. The residue was taken 
up in dilute HCl, the insoluble material filtered off, and the 
filtrate made to 1 00 ml. Aliquots were taken for Ni ( 5 0.0 ml.) and 
Fe (20.0 ml.), and the procedures outlined above were employed. 
The accuracy of these results is lower than in the previous ana- 
lyses due to small sample size. This is indicated in the table by 
the use of fewer significant figures in the reported values. It is 
interesting, however, that the calculated composite analysis based 
on these figures agrees well with corresponding figures obtained 
directly on sample I. 

Oxide composition 

The chemical and X-ray data presented thus far serve to 
characterize the bulk properties of the remnant iron oxides asso- 
ciated with these bronze objects. The optical and electron micro- 
probe work that will be discussed later was done on similar sam- 
ples of oxide. While this work is concerned mainly with the identi- 
fication of metallic inclusions within the oxide matrix, additional 
data on a micro scale will be given on oxide composition and 



^9 



characteristics. These data are consistent with the conclusions that 
are drawn in this section. 

The X-ray diffraction data already mentioned indicate that 
the two remnant materials are mixtures of several oxides of iron. 
These are goethite, lepidocrocite, magnetite, hematite, and maghe- 
mite. The chemical and spectrographic data support this view. 
The spectrographic data also show that from a relatively large 
group of elements none is present in quantities that would be in- 
consistent with oxide derived from the weathering of meteoritic 
material. The small amounts of the elements Si, Ca, Ba, Cu and 
Sn that were found (Table i) can be explained on the basis either 
of condition of burial or the fact that the blade had been fabri- 
cated to a bronze object. The important fact is that the major 
elements present were Fe, Ni and Co, the major elements of iron 
meteorites. 

The data above indicate that the two oxide materials have 
much in common compositionally. Their important difference is 
that sample I, the broad axe blade oxide, is largely magnetite and 
contains appreciably more Ni and Co than the dagger axe oxide. 
It is reasonable to expect that weathering which produces mag- 
netite from a Ni-containing material would retain more Ni than 
weathering that produces goethite and lepidocrocite. Magnetite 
has a 2-valent cation crystallographic site, normally occupied by 
ferrous iron, which is a favorable host position for Ni. Nickel- 
ferrous iron substitutions are well known in mineralogy, the two 
ions having the same plus 2 charge and similar ionic radii. Iron 
oxides with only 3-valent cation positions such as goethite and 
lepidocrocite are known as poor hosts for Ni. Buddhue (1957) has 
summarized data from the meteorite literature along with obser- 
vations of his own that support this point. Iron oxides resulting 
from the weathering of iron meteorites vary greatly in the amount 
of Ni they contain. In some cases the oxide residue has lost essen- 
tially all of the Ni from the parent meteorite. 

The data in Table 3 show that three quarters of the total Ni 
present in sample I is associated with the magnetic fraction. This 
material has been shown by X-ray diffraction to be essentially 



30 



magnetite. The analyses of the dagger axe (34.1 1) oxide (Table 2, 
analyses II, III, IV and V), which is essentially a mixture of 
goethite and lepidocrocite, show that much less Ni has been re- 
tained by these hydrous oxides. A reasonable explanation of the 
observed mineralogy and consequent differing Ni contents in the 
remnant oxides would be slightly different conditions of oxide 
development during burial. The controlling factors in such a case 
might be the effective oxidation potential and acidity. It is, of 
course, not reasonable to assume constant conditions throughout 
the period of oxide development, but this approach may give a 
simplified view of what took place. Magnetite is stable under 
mildly reducing conditions in solutions that are slightly alkaline. 
More oxidizing conditions result in hematite formation over a 
broad range of conditions from mildly acid to strongly alkaline 
(Garrels and Christ, 1965). At a given acidity, a minor change in 
oxidizing conditions could mean the difference between magnetite 
or hematite formation. The specific conditions that lead to mag- 
netite formation are also favorable for Ni retention. Oxidation 
potentials in a burial site would be affected by soil and ground 
water acidity, water table levels, the presence of reducing ma- 
terials such as decaying organic matter, and possibly other factors. 
Significantly different oxidation potentials can be operative over 
distances separated by a few centimeters, or even less. A fairly 
broad range of oxidation conditions could exist within a single 
burial. 



Sampling for optical and electron microprohe examination 

Chips of the remnant oxide were removed from the two 
weapons for optical and electron microprobe examination. The 
samples were mounted in plastic and highly polished surfaces were 
prepared by standard methods. One mount contained 10 frag- 
ments of the broad axe (34.10) blade oxide. Three of these frag- 
ment surfaces are shown in figure i ^. Two mounts were prepared 
from dagger axe (34.1 1) oxide fragments. The single fragment in 
one of these mounts is shown in figure 16. Figure 77 is a macro- 



31 




Fig. 15. — 34.10. Fragments of remnant oxide from metal blade of the broad axe, mounted in plastic and pol- 
ished for microscopic and electron microprobe examination; x 20. Small bright areas are remnant metal. 
Marked areas indicate location of photomicrographs (see figs. 18 and 19). 



Fig. 16. — 34.11. Fragment of remnant oxide from metal blade of the dagger axe, mounted in plastic and pol- 
ished for microscopic and electron microprobe examination; approximately X25. Marked area shows location of 

figure 20. 



graph of the surface of the second mount of dagger axe (34.11) 
material containing five fragments. The mounts were prepared 
in the laboratory of the Mineralogical Institute, University of 
Heidelberg, Heidelberg, Germany, preliminary to optical exami- 
nation by Professor Paul Ramdohr. 

Professor Ramdohr's completely independent examination of 
these oxides using the techniques of reflected light microscopy 
provided an important stimulus to more detailed v^ork. It was 
undertaken at a comparatively early stage in our investigations. 
The chemical and X-ray studies described above were nearing 
completion, and it had become obvious that this work alone would 
not permit a strong case to be built for a specific origin for the 
material from which these oxides were derived. More specific data 
were needed. 



33 



Fig. 17. — 34.11. Fragments of remnant oxide from metal blade of dagger axe, mounted in plastic and polished 
for microscopic and electron microprobe examination; x 15. The lines indicate the path of electron microprobe 

traces. 



In an oral report to one of us (Roy S. Clarke, Jr., March 25, 
1964), Professor Ramdohr summarized his findings based on the 
examination of the polished mounts. He found ample evidence 
to conclude that the broad axe (34.10) blade oxide was very prob- 
ably derived from a meteorite. He observed about 2 percent metal 
in this section, including the three meteoritic minerals kamacite, 
taenite, and some troilite. Kamacite is an alloy of Ni-Fe con- 
taining 4.5 to 7.5 % Ni, while taenite is an alloy of Ni-Fe con- 
taning more than 20 % Ni. Kamacite has a body-centered cubic 
structure and is also known as alpha-iron or ferrite; taenite has a 
face-centered cubic structure and is also known as gamma-iron or 
austenite. Troilite is a meteoritic sulfide (stoichiometric FeS). The 
bulk of the sample was reported to be magnetite and maghemite 
with a little limonite (goethite). The magnetite had the patchy 
appearance of high Ni magnetite, typical of a weathered meteor- 
ite. Ramdohr observed remnant meteorite structures and suggested 
that the presence of taenite lamellae was particularly significant. 
He also concluded from microscopic evidence that the taenite 
lamellae had been hot-worked prior to the onset of weathering of 
the blade. 

Ramdohr's examination of the dagger axe (34.11) oxide did 
not convince him that it was derived from a meteorite. He re- 
ported that it was largely limonite with inclusions of Cu. 

Work to be discussed below has confirmed the validity of 
Ramdohr's identifications and the major conclusion drawn about 
the broad axe (34.10) blade oxide. Additional data obtained by 
microprobe analysis has confirmed the mineral identifications in 
the dagger axe (34.11) point oxide, and allowed us to conclude 
that this oxide, too, may well be meteoritic in origin. 

Electron microprobe studies 

Studies employing an electron microprobe have extended and 
confirmed the chemical and optical investigations described above. 
An electron microprobe is an intricate modern instrument used to 
obtain chemical analyses from small volumes of the surface of a 
solid sample by analysis of emitted X-rays. The possibihty of this 



35 



type of analysis was implicit in Moseley's (19 13) early work on 
X-ray spectra. Castaing (1952) was the first to construct an ana- 
lytical instrument, and he made fundamental contributions to the 
development of the basic theory and philosophy of quantitative 
application. In brief, the instrument focuses a narrow beam of 
electrons on a polished sample surface in vacuum, the beam size 
frequently being as small as one micron (0.001 mm.) in diam- 
eter. Characteristic X-rays are emitted from a small volume of 
sample, the spectra emitted depending on the composition of the 
sample and the energy of the impinging electrons. The wavelengths 
of the emitted X-rays are determined by curved-crystal spectro- 
meters, and the intensities of the X-rays are measured by appro- 
priate counting devices. X-ray wavelengths identify the elements 
present, and specific intensities relate to the amount of a given 
element present. Recent reviews describing quantitative electron- 
probe analysis in detail have been given by Birks (1963) and Keil 
(1967). Archaeological researchers have made good use of the 
electron microprobe in studying specific problems; see, for ex- 
ample, the work of Hornblower (1963), Young (1963), Brill and 
Moll ( 1 9 6 3 ) and Lechtman (in press) . 

The analytical data reported here were obtained in the labo- 
ratory of the Division of Meteorites of the National Museum of 
Natural History, Smithsonian Institution, using a modified elec- 
tron microprobe built by the Applied Research Laboratories, 
Glendale, California. 

Photomicrographs of the areas of the polished sections that 
were analyzed by the microprobe are shown in figures 18-20. The 
locations of these areas are indicated on figures i ^ and 16. The 
photomicrographs were taken on a Vickers Projection Micro- 
scope, an instrument designed primarily for metallographic ap- 
plication. 

Figure / 5 is of an area of broad axe (34.10) blade oxide con- 
taining lamellae of bright metal that appear to be corroded and 
distorted taenite. Taenite is a meteoritic mineral containing Fe, 
fairly large amounts of Ni, and some Co (see above). The matrix 
is a mixture of oxides of varying composition and patchy appear- 



36 




Fig. 18. — 34.10. Photomicrograph (x 230) of a metal-containing area 
of the broad axe oxide (detail of fig. /j). The bright metal is taenite 
and the general structure is suggestive of Widmanstatten pattern. The 
line indicates the path of the electron microprobe trace in figure 21. 



ance. Cracks and a few small holes are also present. Figure is 
an area from a different oxide fragment of this same object. This 
area contains many small blebs of bright metal that appear to be 
corroded kamacite. Similar appearing material is also present in 
the oxide fragment containing the suspected taenite. Kamacite is 
also a meteoritic mineral containing Fe, Ni and Co. The Ni con- 
tent of kamacite is smaller than that of taenite, the maximum Ni 
content being about 7.5 percent. Co content of kamacite tends to 
be higher than in taenite. 



37 



Fig. 19. — 34.10. Photomicrograph (X230) of metal containing area 
of broad axe oxide (detail of fig. i j). The bright metal is kamacite. 
The line indicates the path of the electron microprobe trace in 

figure 22. 



Figure 20 is an area in the oxide from the dagger axe (34.1 1) 
blade; it is one of the few areas in the six mounted fragments of 
this oxide that contains remnant metal of possible meteoritic com- 
position. Metal in other parts of this material was found to con- 
tain major amounts of Cu and was therefore considered to be 
related to the bronze alloy in the immediate environment. The 
area of the photomicrograph contains a number of very small, 
bright areas that were unidentifiable under the microscope. They 
are possibly badly oxidized remnants of kamacite or taenite. This 



38 



Fig. 20. — 34.11. Photomicrograph (X230) of dagger axe oxide (detail 
of fig. 16). The small bright metal areas are not completely resolv- 
able but seem to be of kamacite composition. The letters indicate 
locations of electron microprobe traces in figure 2 j. 



area also contains patchy oxide, a number of small holes, and one 
large crack. The botryoidal development of the oxide along this 
large crack is a typical habit of this material. 

Several spots from each of the areas shown in the photomicro- 
graphs (figs. 18-20) were selected for quantitative microprobe 
analysis. The results are given in Table 4. Analyses i and 2 are 
from the approximate area of figure 79, while analyses 4 through 
7 are from the area of figure 18. Analysis 3 is from the lighter area 
shown at the left corner of the upper particle in /igwre 7^. Analyses 



39 



8 and 9 are from the area of figure 20. They were performed using 
two spectrometers simukaneously. Fe and Ni were determined, fol- 
lowed by Fe and Co from the same spots. Both optical observation 
and total Fe counts were used to establish that all three elements 
were determined on the same spots. The following standards were 
used: 100% Fe, 100% Ni, 100% Co, 5% Ni-95% Fe, and 
50% Ni-50% Fe. Conventional background, absorption and 
fluorescence corrections were made. Several of these same speci- 
men areas were also checked qualitatively for the presence of 
other elements by scanning the emitted X-ray spectrum. Elements 
present in significant amounts with atomic numbers greater than 
II (Na) would be detected by this procedure. Fe, Ni, Co and 
traces of P were the only elements found. A particularly careful 
check was made for the presence of Mn in the metal areas, but it 
was not detected. The only element observed which had not been 
noticed previously is P. It was not sought by wet chemical proce- 
dures, and the spectrographic procedures used are relatively in- 
tensitive to P. On the other hand the electron microprobe is rela- 
tively sensitive to P, and this element is generally present in the 
range of o.i to 0.3 percent in iron meteorites. These findings are 
consistent with the chemistry of meteorite corrosion as discussed 
above. 

The individual analyses in Table 4 strongly suggest that ma- 
terial of both kamacite and taenite compositions are present in the 
broad axe (34.10) blade oxide. Analyses i through 3 are typical 
of meteoritic kamacite. Analyses 4 through 7 suggest taenite, but 
the totals are somewhat low. The absence of other elements of 
atomic number greater than Na and the physical condition of the 
material being analyzed make this a readily explainable result. 
The metal areas are small and undoubtedly thin. The X-ray pro- 
ducing volume of sample undoubtedly contains oxide, but oxygen 
is not detectable by this technique, leading to a low summation. 
The same explanation undoubtedly applies to the metal-con- 
taining area analyses of the dagger axe (34.1 1) oxide. These areas 
are so small that oxide must be included in the probe analysis 
sample. The important observation is that the metal undoubtedly 



40 



TABLE 4 



Electron microprobe analyses of metallic inclusions in 





broad axe 


(J4. iUj and dag 


ger axe (34.11) 


oxides* 




Sample 


Analysis 


Ni 


Fe 


Co 


Total 




No. 


/o 


/o 


/o 


/o 


34.10 


1 


6.9 


92.2 


.6 


99.7 




2 


6.7 


92.6 


.7 


100.0 




3 


6.8 


91.6 


.6 


99.0 




4 


23.0 


65.3 


.3 


88.6 




5 


22.6 


70.0 


.3 


92.9 




6 


27.3 


62.4 


.3 


90.0 




7 


29.3 


65.4 


.2 


94.9 


34.11 


8 


5.2 


86.0 


.4 


91.6 




9 


5.2 


81.5 


.3 


87.0 



* Analyst: Joseph Nelen. 



contains only Fe, Ni and Co in greater than trace amounts, and 
that the ratios of these elements are appropriate to meteoritic 
kamacite. 

A careful examination of other areas of dagger axe (34.11) 
oxide was undertaken with the microprobe. Traces for Fe and Ni 
were made across four fragments as indicated in figure ly. The 
sample was moved under the electron beam at a rate of 96 microns 
per minute, and a distance of approximately 6400 microns was 
traversed. Ni varied from approximately 0.5 percent to a maxi- 
mum of about 4 percent, with Fe varying little from 5 5 percent. 
Ni values generally were in the i to 2 percent range, and only 
infrequently above 2 percent. No additional areas were found 
that seem to contain either remnant taenite or kamacite. 

Simultaneous electron microprobe traces for Ni and Fe were 
also made in three metal-containing areas where quantitative 
probe analysis had been done (see Table 4). Samples were moved 
under the electron beam at a rate of 8 microns per minute. 
Figure 2/ is a reproduction of the trace across the oxide (broad 



41 



axe, 34.10) containing material of taenite composition shown in 
figure 18. This trace started at A in the photomicrograph and 
went to B, and then to C after a small change in direction. Figures 
22 and 2 J are similar traces made in the indicated areas in figures 
79 and 20. These tracings show the change in distribution of Ni 
and Fe as the probe beam moves from oxide, across metal, and 
then back into oxide. 

The Fe and Ni analyses of broad axe (34. i o) blade material of 
taenite composition in Table 4 are confirmed and presented in 
perhaps a more instructive way in figure 21. Four distinct simul- 
taneous peaks in Fe and Ni concentration indicate taenite compo- 
sition, 70 to 76 percent Fe and 26 to 30 percent Ni (peaks at 60 
microns, 85 to 105 microns, 220 to 235 microns and 395 to 405 
microns). These values total approximately 100 percent, excellent 
agreement considering the limitations of this technique. The high 
Ni peaks appeared on the chart recorder as the electron beam was 
optically observed to pass from oxide to bright metal. The oxide 
composition is relatively uniform in most areas, averaging about 
6 percent Ni. The sum of Ni and Fe is about what would be ex- 
pected for a material that is essentially Ni-containing iron oxide. 
Occasional Ni values in the oxide are as high as 1 5 percent and 
correspond generally to lower than average Fe values. This type 
of variation is not surprising considering the patchy appearance of 
the oxide. One Fe peak (at 270 microns) is considerably above the 
average for this element, and it appears to correspond to a slightly 
lower value for Ni. A possible explanation is the presence of a 
small amount of material of kamacite composition just below the 
oxide surface. 

Figure 22 presents the results of a simultaneous trace for Ni 
and Fe across an area containing material of kamacite composition 
indicated in the photomicrograph (fig. 19) of the broad axe (34. 10) 
blade oxide. For convenience of presentation different scales are 
used to present concentrations of the two different elements. The 
main difference is that the Ni data are presented in slightly more 
detail than is the case in figure 21. The metal areas in this case 
contain approximately 7 percent Ni and 92 percent Fe, as would 



43 




O <D CO 1^ CD lO 

lN30a3d 1H9I3M 



be expected from the data in Table 4. The composition of the ox- 
ide is similar to that discussed above, with the exception that the 
average Ni content is probably somewhat lower. 

Figure 2j presents the results of three simultaneous traces for 
Ni and Fe from the areas indicated in figure 20, the photomicro- 
graph of the dagger axe (34.1 1) point oxide. Each of these areas 
contains a small amount of material that closely approximates 
kamacite composition and confirms similar data in Table 4. It is 
obvious that pieces of remnant metal are present only in very 
small amounts on this surface. The oxide composition is similar to 
that of the broad axe (34.10) blade material discussed above. 

Octahedrite structure and remnant oxide 

The analytical and other data reported above is more mean- 
ingful when viewed in the context of modern work on the struc- 
ture of octahedrite meteorites. This background material is im- 
portant to a valid interpretation of our data. The compositions 
and remnant structures observed in the case of the broad axe 
(34.10) blade material are consistent with metallurgically worked 
meteoritic material followed by weathering during burial. 

Iron meteorites are classified on the basis of their crystallo- 
graphic structure. This structure is displayed when flat, highly 
polished surfaces are etched with appropriate chemical reagents. 
Specimens of the largest single group of iron meteorites, called 
octahedrites, develop a pronounced pattern under these conditions 
which is referred to as a Widmanstatten pattern after Alois Wid- 
manstatten who observed the phenomenon in 1808. The Widman- 
statten pattern of a coarse structured octahedrite meteorite is 
shown in figure 24. This type of pattern, sufficiently coarse to be 
obvious to the unaided eye, is known only in iron meteorites. Simi- 
lar patterns have been observed in artificial materials, but they 
require high magnification to be seen. Excellent reviews of the dis- 
covery and early history of the Widmanstatten pattern have been 
given by Smith (i960, 1962) and Paneth (i960). A general intro- 
duction to iron meteorites and their classification has been given 



46 



Fig. 24. — Widmanstatten pattern revealed in the Arispe, Mexico, iron meteorite. The lamellae in the photo 
graph are kamacite, their differing reflectivities are caused by differing crystallographic orientations. The kama 
cite plates are separated by taenite lamellae (see fig. 26) that are too fine to be revealed at this low magnifica 
tion, NMNH 229, polished and etched slice (about xo.4). 



by Mason ( 1962), and a detailed discussion of their metallography 
by Perry (1944), Buchwald (1966) and Axon (1968). 

The Widmanstatten structure contains important clues for an 
understanding of the history and origin of iron meteorites, and for 
this reason has been extensively studied over the years. Recent 
work relying heavily on modern metallurgical concepts and elec- 
tron microprobe phase analysis has confirmed and extended earlier 
ideas based on less detailed data. The broad outlines of the process 
of Widmanstatten pattern formation are now generally agreed 
upon. Important recent papers in this field are by Feller-Kniep- 
meier and Uhlig (1961), Agrcll, Long and Ogilvie (1963), Gold- 



47 



stein and Ogil vie (1965), Short and Andersen (1965), Reed (1965) 
and Buchwald (1967). 

The observed structure of octahedrite meteorites may best be 
understood in terms of a slowly cooled Fe-Ni alloy containing at 
least 6.5 percent Ni. The equilibrium phase relationships involved 
below 1000° are approximated in figure 2^. This diagram, taken 
from Goldstein and Ogilvie (1965), is based on the data of Owen 
and Liu (1949). It may be used to explain in broad outline the de- 
velopment of the Widmanstatten pattern. An example of this pro- 
cess is an alloy of Fe containing 7 percent Ni which is cooled 
slowly through the sub-solidus region of taenite stability. Cooling 
rates of meteorites are slow enough to permit the development of 
large single crystals of taenite, and this crystal structure controls 
the subsequent development of the Widmanstatten pattern. At 
about 750 C the cooling alloy enters the taenite-kamacite region 
of stability and subsequently kamacite nucleates and grows at the 
expense of taenite. The transformation of taenite to kamacite is a 
diffusion-controlled process. The phase relationship is such that 
as the temperature drops both taenite and kamacite become richer 
in Ni. Around 450 °C the diffusion rate of Ni in Fe becomes so 
low that the observed Widmanstatten pattern is frozen in. Ni 
concentration gradients are present in the structure due to arrest- 
ed diffusion resulting in non-attainment of equilibrium. 

Ni gradients are particularly pronounced in taenite at the 
taenite-kamacite interface (Short and Andersen, 1965). Figure 26 
is a photomicrograph of a small area of Widmanstatten pattern 
from a sample of the Campo del Cielo, Argentina, meteorite, one 
of many octahedrite meteorites that could have been selected to 
make this point (see Cassidy et aL, 1 9 6 5 , for an introduction to the 
Campo del Cielo, Argentina, strewn field and the many meteorite 
specimens that have been recovered there over a period of nearly 
400 years). The light grey matrix is kamacite, and the three par- 
allel lamellae are taenite. The presence of distinct, sharp borders 
at the taenite-kamacite interface corresponds to high Ni concen- 
tration. This photomicrograph is at the same magnification as the 
oxide photomicrographs shown earlier. 



48 



Fig. 26. — Photomicrograph (X230) of a selected area in a section 
of the Campo del Cielo, Argentina, meteorite (NMNH-2253). The 
gray background is kamacite, the three vertical lamallae are taenite. 
The letters A and B indicate the path of the electron microprobe 
trace in figure ij. 



Ni and Fe electron microprobe profiles {jig. 27) were obtained 
on the Campo del Cielo section {jig. 26) by the same electron 
microprobe technique used earlier for the weapon oxides. The 
M-shaped Ni profiles for taenite expected on the basis of careful 
work by others (for instance, Short and Andersen, 1965) were 
reproduced in an approximate form. The diffusion gradients at 
the taenite borders are obvious, although not as sharp and well 
resolved as would be the case with more painstaking work. 



50 



Figure ly summarizes in a general way the Ni and Fe distribu- 
tion over a portion of the Widmanstatten pattern of an octa- 
hedrite meteorite. The areas of high Ni content are traces across 
taenite, and the highest Ni values are at the taenite-kamacite 
borders. The Ni gradient within a given taenite is greater with 
increasing width of the taenite lamellae. The three taenites shown 
vary from less than lo microns to approximately 30 microns in 
width. These are well within the normal size range for the feature. 
The bulk of the material represented by the trace in figure zy Is 
kamacite. The Ni concentration in kamacite is generally uniform 
but does decrease measureably near taenite lamellae. For both 
kamacite and taenite, the Fe distribution varies inversely with Ni. 
Fe and Ni total approximately 100 percent, a small amount of 
Co being present in both phases. 

Figure ly may also be used to correlate data from figures 21 
to 2j and to relate these data to measurements made on an unox- 
idized specimen of the Campo del Cielo meteorite. Figure 21 is 
suggestive of figure 2y, both figures having three restricted areas 
of high Ni content. These peaks represent the metal areas of figure 
18, and the Ni and Fe distribution in figure 21 is roughly that of 
the known taenite lamellae in figure 2y. The significant difference 
between figures 21 and 27 is the absence of diffusion borders in 
figure 21. Diffusion borders are represented in figure 2y by the 
high Ni spikes on either side of the taenite peaks at the kamacite- 
taenite interfaces. However, the gross composition of the lamellae 
in our weapon oxide sample, their widths and relative spacing, 
and their presence in a Ni-containing iron oxide strongly suggest 
that they are remnant taenite lamellae derived from an octahedrite 
meteorite. 

Figures 22 and 2 j contain regions that are similar in composi- 
tion to the kamacite regions of figure 2y and to the more accurate 
analyses reported in Table 4. Figure 22 contains three areas where 
both Ni and Fe content is relatively uniform. The Ni is in the 6 to 
7 percent range and Fe is 92-93 percent. The small metallic parti- 
cles analyzed in the areas represented by figure 2 j give the same 
general picture as that presented in Table 4. The left-hand set of 



51 




lN30d3d 1H9I3M 



curves is particularly similar to the data presented in figure 22. 
The high iron peaks in the other two sets of data in figure 2 j sug- 
gest kamacite composition. Each is too high for an oxide composi- 
tion, and is accompanied by an appropriate amount of Ni for 
kamacite. 

Rounding of the apparent taenite peaks in figure 21 probably 
requires an explanation other than simply corrosion. Data similar 
to that presented in figure 2 j was taken from a second Campo 
del Cielo meteorite specimen. One containing areas of terrestrial 
weathering was selected, and data was taken over a taenite lamella 
embedded in oxide. The Ni and Fe distribution curves in the oxide 
were similar to the oxide curves discussed previously. Within the 
taenite lamella, however, the Ni diffusion borders were still pres- 
ent. This would seem reasonable on the basis that metal containing 
the highest Ni would be expected to be the most resistant to weath- 
ering. 

A reasonable explanation for the lack of diffusion borders 
would be metallurgical working of the sample. Ramdohr (see 
above) reported that he found evidence for this in his optical ex- 
amination of the taenite lamellae. It is well known that heat treat- 
ment and working damage the Widmanstatten structure of mete- 
orites. The dissolution of taenite borders is one of the effects 
noted (Wood, 1964, p. 436). Buchwald (1967) states that "we 
know now that the diffusion in the metallic matrix is so rapid that 
significant transformations take place already during short heat- 
treatments above 5 00-600 °C. " At higher temperatures both kama- 
cite and taenite are brought into the taenite region of the phase 
diagram {fig. 2^). On rapid cooling of this material a new phase, 
alpha2, is formed. It may well be that the materials of kamacite 
and taenite composition that we are dealing with in this study 
have undergone a history of heat treatment and mechanical work- 
ing and have been transformed to the alpha^ phase while main- 
taining remnant structures and compositions. A history of this 
type would have a rounding effect on the Ni profile in remnant 
taenite lamellae. In the discussion of the fabrication of these ob- 
jects it was pointed out that the bronze members were cast on to 



53 



the shaped blades. This technique would have required heating of 
the iron blades to temperatures above 600 °C for a few minutes. 

A further observation that is consistent with a history of marked 
transformation of meteoritic material was the lack of fine-struc- 
ture in the metallic particles. Both unaltered meteoritic kamacite 
and taenite typically contain fine-structure, which develops on 
etching and is observable under the microscope. The structure 
within the taenite in figure 26 is an example of this. An attempt 
was made to etch the metal remnants in the broad axe (34.10) 
blade oxide, but no change was observed even with long exposure 
to a strong Nital etching solution. 

Remnant metal in Wolf Creek meteoritic oxide 

The literature contains several reports on remnant metallic 
inclusions in known meteoritic oxide from the large Wolf Creek 
Crater, Western Australia (Cassidy, 1954). Our interpretation of 
these data leads us to conclude that this oxide is analogous in im- 
portant ways to the broad axe and dagger axe oxide we have 
studied. Wolf Creek meteoritic oxide is relatively abundant and 
until recently was the only meteoritic material recovered from this 
locality. La Paz (1954) examined large polished surfaces of two 
Wolf Creek specimens and reported that the oxide contained 
"small granules and sinuous veins of metallic nickel-iron." These 
were clearly revealed although not abundant. White, Henderson 
and Mason (1967) have also reported sparsely disseminated me- 
tallic particles in samples of Wolf Creek oxide. One of these parti- 
cles was analyzed by the electron microprobe and found to con- 
tain 21.3 percent nickel, a value that was attributed to selective 
enrichment during weathering. The possibility that the particle 
was remnant taenite, an explanation that seems more plausible, 
apparently was not considered. Knox (1967) also reports the pres- 
ence of small metal particles surviving in this material, one of 
which he identified as kamacite. E. P. Henderson (personal com- 
munication, 1968) has examined Wolf Creek oxide and reports 
finding areas where the oxide structure clearly reveals remnant 



54 



Widmanstatten pattern. All of these observations are consistent 
with derivation of Wolf Creek oxide from an octahedrite meteor- 
ite, a meteorite with a Widmanstatten pattern of kamacite and 
taenite. Furthermore, they support the suggestion by Taylor 
(1965) that unoxidized octahedrite meteorite specimens found 
near the Wolf Creek Crater may be part of the crater-forming 
body. The fragmented nature of the recovered specimens is also 
suggestive of material found at known octahedrite impact sites 
such as Barringer Meteor Crater, Arizona; Henbury Craters, 
Australia; Wabar Craters, Arabia; and Sikhote-Alin Craters, 
eastern Siberia (Hey, 1966). Another similarity worth noting is 
that both completely oxidized remnant material and essentially 
unoxidized specimen material are abundant at Barringer Meteor 
Crater, Arizona (Nininger, 1956). 

Native iron 

Iron and nickel-iron are known as rare minerals from a small 
number of natural occurrences. The iron-bearing basalts of Disko 
Island, Greenland, are the best-known example (Pauly, 1969). 
Several reasonably large masses of native iron have been recovered 
from this area, and iron disseminated in the basaltic matrix is com- 
mon. The total amount of iron, however, is very small and the ge- 
ographic area involved limited. This type of iron deposit has never 
been a major source of the metal, and reports of its use in utensils 
are doubtful. 

Native iron specimens normally contain Ni and small amounts 
of Co, S, P and C. These are the same elements associated in iron 
meteorites but there are marked differences in composition, min- 
eralogy and petrography. The Ni in native iron is generally in the 
2 percent range and rarely over 3 percent. Native nickel-iron 
specimens containing major quantities of Ni are also known, their 
Ni contents normally ranging above 60 percent. Terrestrial iron 
or nickel-iron is essentially unknown in the composition range of 
kamacite and taenite from iron meteorites. The Widmanstatten 
pattern of iron meteorites, with its intimate association of kama- 



55 



cite and taenite, has not been reliably demonstrated in terrestrial 
irons and probably does not exist there. Native iron structures and 
mineral associations are more suggestive of white cast iron and 
slags than they are of meteorites. It is highly unlikely that native 
irons were used to manufacture the blades of the weapons that 
produced the remnant oxide studied here. 

Meteoritic origin of remnant oxides 

The data in the preceding sections allow us to conclude that 
the iron blades associated with these bronze objects were fabri- 
cated from meteoritic iron. Undoubtedly, the meteoritic structures 
originally present were damaged both in fabrication of the blade 
and in the casting of the bronze tangs. The two blades may have 
come from different iron meteorites, the data being ambiguous on 
this point. In any event, the objects had sufficiently different 
weathering histories, undoubtedly as a result of conditions during 
burial, to produce remnant oxides of markedly different charac- 
ter. The broad axe (34.10) blade oxide developed under suffi- 
ciently reducing conditions to produce a corrosion product largely 
of magnetite and therefore to maintain the bulk of the Ni original- 
ly present. The dagger axe (34.1 1) point oxide was formed under 
more oxidizing conditions and a limonitic oxide developed, most 
of the original Ni being lost. Remnant structures observable in the 
oxide of the broad axe (34.10) blade and the presence of material 
of kamacite and taenite composition strongly suggest that this ox- 
ide was derived from an octahedrite meteorite, a meteorite show- 
ing a definite Widmanstatten pattern. The presence of metal of 
kamacite composition in the dagger axe (34.1 1) oxide is evidence 
for weathering of an iron meteorite. In this case, however, neither 
structural nor compositional data sufficient to establish the struc- 
tural class of the meteorite was developed. We find much in our 
data to support the general thesis of a meteorite being the source 
of the iron and nothing that is inconsistent with it. 



56 



V. SIGNIFICANCE OF THE FINDINGS 



What significance do our findings have in the context of the 
metalworking technology of early Chou China? We have proven 
to the limits of our present ability the hypothesis of the meteoritic 
origin of the iron in these objects, and we have also advanced 
evidence for casting-on joining in these two weapons. For these 
facts to be relevant, they must be related to other iron, bronze 
and composite objects both from China and from other cultures. 
Before these relationships can be pointed out, we must place the 
weapons in their correct chronological and geographical locations. 

With the two weapons came a statement in Chinese by Ch'u 
Te-i which has been translated as follows (Freer Gallery of Art, 
1946, p.91): 

"Twelve bronze weapons of the Chou dynasty. In the 6th 
moon of the 20th year of the Republic of China [193 1] a na- 
tive dug up in Wei-hui-fu, Honan Province, not far from An- 
yang District, a group of ancient weapons, namely: six hal- 
berds [ko], one lance [mou\, two hatchets [fu\, and three 
knives [tao]. On one of the knives are the two characters 
K'ang Hou [Marquis K'ang]. K'ang-shu was a younger 
brother of King Ch'eng of Chou, and his appanage included 
what are now An-yang District and Wei-hui Prefecture. 
[K'ang-shu was actually Ch'eng Wang's uncle, as pointed 
out in footnote 12 a, ibid,'] These 1 2 weapons are undoubted- 
ly relics of the Chou Marquis K'ang. Among them is a Ch'ih- 
yu knife, a circular knife, an ox-head halberd and a halberd 
inlaid with shell. Antiquities so remarkable in form and make 
have not been seen hitherto. By students of the manners and 
customs of ancient Chou, they must be regarded as great 
treasures. Recorded by Ch'u Te-i. The nth day of the loth 
moon of the 20th year of the Republic of China [ 1 9 3 1 ] • " 
Mr. Lodge commented as follows: 

"Ch'u Te-i is said to have been associated in some capacity 
with the famous collections gathered and published by the late 



57 



Tuan Fang. In any case, his record of these weapons is at 
present our only available information as to where and when 
they were found, and there is no apparent reason to doubt its 
essential accuracy. Just how Ch'u groups the weapons as ko, 
mou^ and fu is not clear in every case; but those he describes 
as *a Ch'ih-yu knife' (34.3), 'a circular knife' (34.4), 'an ox- 
head Halberd' (34.7), 'a halberd inlaid with shell' (34.8), are 
all easily identifiable." 

To recapitulate, Ch'u says that these weapons were included in 
a group of twelve found together in the ground by a native (prob- 
ably a farmer) in June, 193 1. He says that the find occurred near 
An-yang in the Wei-hui prefecture of the Province of Honan. (A 
year later, controlled excavations started near this area, at a 
place called Hsin-ts'un.) While this information does not have the 
authority of a report from a controlled excavation, it does seem 
to be accurate. Ch'u Te-i, as implied in Mr. Lodge's comments 
above, had a great deal of experience with ancient Chinese bron- 
zes, and he wrote the C. T. Loo Catalogue of 1924 along with 
other books on Chinese antiquities (TaT'ang, 1962). 

On the other hand, Ch'u's statement is not without ambigu- 
ity. We must remember that this information accompanied the 
weapons when they were purchased and must have added to their 
market value; the statement also consists of hearsay evidence at 
best. What can we find to confirm the assertions set forth by 
Ch'u? 

Clandestine digging is known to have been in progress in the 
area of the Hsin-ts'un cemetery and at other locations within the 
boundaries of the old Wei state during 193 1, the date that Ch'u 
assigns to the finding of these weapons. "Before 1932 the region 
[Hsin-ts'un] was a happy hunting ground for curio-diggers. Plun- 
dering was going on with great enthusiasm in the spring of 193 1 
when the provincial government prohibited digging, after which 
official excavations were carried out for four successive seasons in 
1932-33" (Cheng, p. 76). A group of eight objects in the Royal 
Ontario Museum is said to have come from Hsin-ts'un by way of 

58 



the art market; one chi (R. O. M. number NB. 3155) is inscribed 
with the character for Hou or marquis. W. C.White says (1956, 
p. 164) that when objects started to be unearthed at Hsin-ts'un in 
the spring of 193 1, "At once dealers from the port cities swarmed 
to the area, paying exhorbitant prices for anything they could lay 
their hands on " The demand thus created would have encour- 
aged further clandestine digging, and the rapid movement of these 
objects onto the art market would have further obscured the exact 
facts of any particular case. 

The connection of our weapons with other objects may be the 
surest way of confirming Ch'u's statement. We have attempted 
below to draw parallels for our ko and chi weapons with those 
excavated by Kuo Pao-chiin at Hsin-ts'un, Hsiin Hsien. Ch'en 
Meng-chia (1948 and 1955) groups our large tao together with 7 
other weapons and vessels, including the Malcolm kuei (see be- 
low). All of these bear a K' ang-hou inscription. He says that this 
set was probably found together, in the same pit, at the same time, 
but it is clear that his reason for grouping them together is the in- 
scription. He does not, therefore, mention the uninscribed weapons 
that we possess. Ch'en also states the following: that the objects 
were not all made at the same time; that the Malcolm kuei was 
probably made by a Shang craftsman; and that the location of 
the finding of the set is not precisely known. He mentions three 
alternative possibilities for the provenance: Wei-hui-fu (modern 
Chi Hsien), Chiin Hsien (or Hsiin Hsien), and Ku-yii-ts'un in 
Hui Hsien. All three lie within a fifty-mile radius to the south of 
An-yang. While Ch'en does not advance actual evidence for any 
of the three, he believes that the Hui Hsien site is the most prob- 
able. Because of the clandestine nature of the finding of these ob- 
jects and the intervening years, we do not think that we can do 
more than accept what Ch'u says, and assume that these weapons 
were found near Wei-hui-fu. At least we can be fairly sure that 
they were found within the territory ruled by Marquis K'ang, 
Duke of Wei. The assertion that they were all found together also 
stands, there being no evidence to the contrary. We can infer that 
they were also buried together. 



60 



TABLE 5b 

Spectrographic analyses of the set of bronze weapons. f 



Bi Zn Cr Mg As Mn Si 

0/ 0/ 0/ 0/ 0/ 0/ 0/ 

/o /o /o /o /o /o /o 



Ca 



Al FGA 
% No. 



.2 nd <.001 <.001 .03 <.001 .02 .1 .001 34.3 



nd nd nd .0005 .01 .0005 .03 .005 <.0005 34.4 



.06 nd nd .001 .1 .0005 .03 .003 <.0005 34.5 



nd nd <.001 <.001 .03 <.001 .001 nd .001 34.6 



.03 nd nd .001 .2 .0005 .005 .0007 <.0005 34.7 



.005 nd nd .0007 .1 .0005 .0005 <.0005 <.0O05 34.8 



nd nd nd .0007 .1 .0005 .001 .005 <.0005 34.9 



nd <.001 <.001 nd 



0.01 
nd to 
0.1 



— nd 34.10 



TABLE 5a 

Wet chemical analyses of the set of bron2e weapons.* 



TABLE 5b 

Spectrographic analyses of the set of bronze weapons.f 



FGA 
No. 


typet 


place 
sampled 


Cu 


Sn 


Pb 

% 


Fe 

% 


Zn 


Total 

% 


Date 
Analyst 

Size of 
samples 


Au 


Pb 

% 


Ag 

% 


Fe 

% 


Co 


Ni 

% 


Sb 

% 


Bi 

% 


Zn 

% 


Cr 

% 


Mg 

% 


As 
% 


Mn 

% 


Si 


Ca 

% 


Al 

% 


FGA 
No. 


34.3 


- 


handle 


83.66 § 
83.4 


13.48 
14.52 
14.0 


0.0 
0.0 
0.0 


- 


- 


97.4 


6/18/62 
EWF 
80 mg. 


- 


.03 


.03 


.009 


.009 


.03 


.01 


.2 


nd 


<.001 


<.001 


.03 


<.001 


.02 




.001 


34.3 


34.4 


ch'i 


bottom 


79.64 
83.37 
81.5 


17.16 
16.17 
16.7 


0.0 
0.15 
0.1 


0.0 
0.0 
0.0 


0.0 
0.0 
0.0 


98.3 


5/26/69 

IVB 
56 mg. 


nd 


.01 


.01 


.04 


,005 


.001 


.001 


nd 


nd 


nd 


.0005 


.01 


.0005 


.03 


.005 


<.0005 


34.4 


34.5 


. 


hafting 
ridge 


87.20 
87.34 
87.3 


10.92 
11.37 
11.2 


0.47 
0.46 
0.5 


0.0 
0.0 
0.0 


0.0 
0.0 
0.0 


99.0 


6/2/69 
IVB 
56 mg. 


<.005 


.02 


.08 


.06 


.007 


.06 


.01 


.06 


nd 


nd 


.001 


• 


.0005 


.03 


.003 


<.0005 


34.5 


34.6 


tao 


bottom 


85.80 
86.54 
86.2 


11.10 
11.64 
11.4 


0.0 
0.0 
0.0 


- 


- 


97.6 


Ill 


- 


.03 


.09 


■> 


.009 


.07 


.01 


nd 


nd 


<.001 


<.001 


.03 


<.O01 


.001 


nd 


.001 


34.6 


34.7 


ko 


top of tang 


87.24 
87.10 
87.2 


6.15 
6.00 
6.1 


4.14 
4.60 
4.4 


0.0 
0.0 
0.0 


0.0 
0.0 
0.0 


97.7 


5/28/69 

IVB 
56 mg. 


.005 


.» 


.05 


.01 


.007 


.03 


.15 


.03 


nd 


nd 


.001 


.2 


.0005 


.005 


.0007 


<.0005 


34.7 


34.8 


ko 


top of haft 


83.07 
83.07 
83.1 


14.87 
14.59 
14.7 


1.98 
1.98 
1.98 


0.0 
0.0 
0.0 


0.0 
0.0 
0.0 


99.8 


6/3/69 
IVB 
56 mg. 


<.005 


.6 


.05 


.06 


.007 


.005 


.02 


.005 


nd 


nd 


.0007 


■ 


.0005 


.0005 


<.0005 


<.0005 


34.8 


34.9 


ko 


break 
in blade 


85.80 
85.67 
85.7 


12.99 
12.69 
12.8 


0.0 
0.15 
0.1 


0.0 
0.0 
0.0 


0.0 
0.0 
0.0 


98.6 


5/22/69 

IVB 
56 mg. 


<.005 


.005 


.01 


.01 


.007 


.03 


.04 


nd 


nd 


nd 


.0007 




.0005 


.001 


.005 


<.0005 


34.9 


34.10 


ch'i 


tang 


81.91 
81.77 
81.8 


15.78 
15.71 
15.8 


tr. 






97.6 


12/22/60 
EWF 
75 mg. 




0.01 
0.1 


0.001 
to 
0.01 


0.01 
0.1 


0.001 
0.01 


0.01 
0.1 


0.01 
0.1 




nd 


<.001 


<.001 


nd 


nd 


0.01 
0.1 




nd 


34.10 



TABLE 5a continued 



TABLE 5b continued 



FGA 
No. 


type 


sampled 


% 


Sn 

% 


Pb 

% 


Fe 


Zn 

% 


Total 

% 


Date 
Analyst 
Size of 
samples 


Au 

% 


Pb 

% 


H 
% 


Fe 

% 


Co 

% 


Ni 


Sb 

% 


Bi 

% 


Zn 

% 


Cr 

% 


Mg 
% 


As 
% 


Mn 

% 


Si 

% 


Ca 
% 


Al 

% 


FGA 
No. 


34.10 


ch'i 


back of 
blade 


81.85 
81.55 
81.7 


14.55 
14.60 
14.6 


0.0 
0.0 
0.0 


- 


- 


96.3 


4/14/64 

IVB 
100 mg. 


<.005 


.01 


.001 


.001 


.005 


.005 


.001 


nd 


nd 


nd 


.0007 


.01 


.0005 


.0005 


.0005 


<.0005 


34.10 


34.11 


ko 


tang 


86.34 
84.55 
85.4 


12.03 
12.36 
12.2 


2.05 
2.19 
2.2 


- 




99.8 


Ill 


— 


— 


to 
1.0 


0.1 


0.1 


0.1 


to 
0.1 


— 


nd 


<.001 


<.001 


to 
1.0 


nd 


0.01 




1.0 


34.11 


34.12 


small 
blade 


top 

back of 
blade 


85.47 
85.5 


10.2 
10.2 


2.07 


- 


~ 


97.8 


6/3/69 
IVB 
50 mg. 


<.005 


.45 


.05 


.005 


.005 


.01 


.02 


nd 


nd 


nd 


.003 


.08 


.0005 


.01 


.01 


.001 


34.12 


34.13 


blade 


top of 
blade 


90.28 
90.31 
90.3 


7.25 
7.50 
7.4 


0.92 
0.91 
0.9 


0.0 
0.0 
0.0 


0.0 
0.0 
0.0 


98.6 


5/27/69 

IVB 
56 mg. 


.005 


.35 


.02 


.06 


.005 


.03 


.04 


nd 


nd 


nd 


.03 


.15 


.001 


.3 


<.0005 


<.O0O5 


34.13 


34.14 


chi 


bottom 


85.69 
85.88 
85.8 


12.94 
12.78 
12.9 


0.31 
0.15 
0.2 


0.0 
0.0 
0.0 


0.0 
0.0 
0.0 


98,9 


6/20/69 
IVB 
56 mg. 


nd 


.22 


.16 


.05 


.002 


.01 


.22 


.07 


nd 






.22 










34.14 



Notes to tables 5a and 5b : 

* The wet analyses were done by the same method as those in Gettens, 
1969 (ch.III). EWF = EUsabeth West FitzHugh. IVB = Ilona V. Bene. 

f Since the spectrographic analyses were done at different times and 
places and with different experimental conditions, they are included 
here simply for interest and not for precise use; therefore, dates, sample 
sizes, etc. have been left out of table 5b (see also Gettens, 1969, table IB). 
The analyses of 34.4, 34.5, 34.7, 34.8, 34.9, 34.10 (back of blade), 34.12 
and 34.13 were all run by Harold Westley of the C.A.L., U.S.N.M., in 
July, 1970. 



t The type names used here are taken from the old Freer bronze catalogue 
(Freer Gallery of Art, 1946). 

§ In all cases except for 34.12, the analyses consist of two duplicate 
analyses and an average. The sample taken was spUt into two portions, 
visible contamination and corrosion products picked out under a bino- 
cular microscope, and two duplicate analyses run. In all cases, the average 
has been rounded to one figure beyond the decimal point. In the case of 
34.12, only one analysis was available, so this is given again as the 



62 



TABLE 5b continued 



Zn Cr Mg As Mn Si Ca Al FGA 
% % % % % % % % No. 



nd nd .0007 .01 .0005 .0005 .0005 <.0005 34.10 



0.1 



1.0 



0.001 



nd <.001 <.001 to nd to 



0.01 



0.1 

to 34.11 
1.0 



nd nd .003 .08 .0005 .01 .01 .001 34.12 



nd nd .03 .15 .001 .3 <.0005 <.0005 34.13 



nd 



.22 



34.14 



Viewing these weapons as a group, however, it does not seem 
Hkely that they were made simultaneously or in the same work- 
shop {fig. 28). The chemical analyses of these weapons make this 
idea even more believable (Tables 5 a and 5 b). Take for instance 
the blade, or as Loehr (1956, p. 8) calls it, a socketed axe, 34.13. 
This contains more Cu than any of the others in the group, some 
Sn, and a little Pb, and the analysis differs greatly from any of 
the other weapons. On the other hand, weapons which we would 
think of as typically Hsin-ts'un forms such as 34.5, 34.7, ^.nd 
34.14 contain about 86 to 87 percent Cu; 34.5 and 34.14 are 
very similar in their analyses, containing 11 or 12 percent of Sn 
and a little Pb, while 34.7 has much higher Pb content than these 
two. The analyses of 34.10 and 34.1 1 do differ to some extent. At 
this stage it is difficult to attribute these differences to their actu- 
al causes; the weapons may have come from different workshops, 
different ores may have been used, a development of technique 
may have taken place in a single workshop, or the bronze found- 
ry practice may have been sufficiently non-uniform to show these 
variations as a consequence. Some technical analyses of other ma- 
terial from the same site (now at the Academia Sinica on Taiwan) 
might shed some light on this matter. We hope that this table can 
serve as a base for consideration of other early Chou bronze 
weapons, but we cannot draw any definite conclusions from it at 
this stage. 

The typological viewpoint promises to be of more help. 
Judging from the different styles of ko represented in this collec- 
tion, one may assume that the weapons were made at different 
times or in different workshops. (Compare 34.5, 34.8, 34.9, 34.7, 
and 34.11 in fig. 28; see also the old Freer Catalogue, pi. 48 and 

49') 

Sinologists do not yet agree on the typological evolution of the 
ko. The most intensive study of the typology of the ko was writ- 
ten by James Menzies, but he only follows the evolution of the ko 
down to the end of the Shang dynasty. Li Chi has also studied 
the ko in detail. Magdalene von Dewall has traced the evolution 
of the ko at the Hsin-ts'un site. She remarks that our set contains 



63 



Fig. 29 a. — Broken blade of the type ch'i, FGA 34.14. Length, 21.3 cm. 




I'iG. 29 b. — Broken blade of the type c7//, VGA 34.14. 



ko which can be classed under Li Chi's class PV (34.1 1 - the iron- 
bronze ko), PVI (34.7, 34.8, 34.9) and the intermediate form be- 
tween PVII and PVIII. From the inscription ''K'ang hou'' on 
the large slashing knife or tao^ she links our set with other inscrib- 
ed objects from this cemetery, and she concludes: 

"It is clear that both phases must have been so close in time 
that the same historical personage could be mentioned, to wit 
as contemporary, in both instances. The span of two successive 
generations is thus, indeed, the longest period for which this 
fact makes allowance. In terms of historical chronology, this 
span must have covered the era immediately following the 
Chou conquest, dominated by the strife for military and po- 
litical consideration in which the Prince of Wei is known to 
have taken an active part" (p. 527). 

It seems that these weapons might have come from the tomb 
of the Prince of Wei which must have been in the Hsiin Hsien 
area and was looted. This possibility appears even more probable 
when we consider that many of these weapons have correspon- 
dences in the scientifically excavated material. Compare, for ex- 
ample, our 34.7 with M 42: 103 (Kuo, 1964, pi. XVIII no. 15) 
and our 34.8 with M 42: 14 (Kuo, 1964, pi. XIX no. 13). The 
larger and more ornate weapons which do not correspond to the 
excavated material add weight to the hypothesis of princely own- 
ership of these weapons. 

A hitherto unpublished weapon from this group contributes 
further evidence to the argument. It is a thinly-patinated blade of 
the type chi (our 34.14), with both the top, back-curving portion 
and the blade broken off {fig, 2^ a and b). A perforation appears 
where the base of the blade would be, as on M 42 : 108 (Kuo, 1964, 
pi. XXIII no. i) and the lashing holes appear in the same places 
as they do on M 42 : 102 (Kuo, 1964, pi. XXIII no. 2, pi. LXVIII 
no. i). We originally thought that this object might be spurious, 
but it is too close to the excavated material in style to discard as 
a fake, and its chemical composition also fits in closely with the 
rest of our group. The inscription also has some interesting fea- 
tures visible even to a non-epigrapher. One character appears on 



66 



the side which shows when the blade points to the left. This might 
possibly be a corruption of hou as written on the Freer tao (Freer 
Gallery of Art, 1946, p. 94). On the other side of the blade four 
characters appear. These might read "heaven king's son" with 
the fourth character uncertain. This formula could apply to Mar- 
quis K'ang, the Prince of Wei, since he was the brother of King 
Wu. The elucidation of this inscription and in fact a detailed dis- 
cussion of all of the weapons must wait for a qualified epigrapher 
to consider the problem. 

A surprisingly large amount of material exists for this study. 
For instance, the Malcolm Kuei described by Yetts (1937) and 
Ch'en (1948 and 1955) contains an inscription which mentions 
Marquis K'ang. The script style is very similar to that on our tao 
34.6. The kuei also has a well-preserved surface, bronze in color, 
similar to the silvery surface on some of our weapons, especially 
34.3 and 34.4. A weapon mentioned by Loehr (1956, no. 105) is 
of the same type as our 34.6 and is also inscribed. Loehr connects 
this with our t^o, but considers that it is a Shang prototype of 
our weapon. 

Of greatest interest to the historian is the presence of textual 
evidence for the existence of Marquis K'ang. This evidence is dealt 
with by Yetts and Ch'en in some detail. K'ang is the recipient of 
"the announcement to the Prince of K'ang" in the Shu ching 
(Legge, vol. Ill, p. 38 iff.). Here the Duke of Chou speaks as 
Regent with the authority of King Ch'eng and charges K'ang 
with the careful and conscientious management of the area which 
previously contained the Shang capital. It is of more interest to 
us that K'ang, later Marquis Wei, was also present at King 
Ch'eng's funeral in 1004 B.C. (Loehr, 1968, pp. 14 and 17), and 
heard "the testamentary charge" {Shu ching, book XXII; Legge, 
vol. Ill, p. 544 ff.). From this funeral we can derive some idea 
of what the funeral of Marquis K'ang himself must have been like. 
"The salvage men set out the screens ornamented with fig- 
ures of axes, and the tents. Between the window and the door, 
facing the south, they placed the different mats of bamboo 
basket-work, with their striped borders of white and black 



67 



silk; and the usual bench and adorned with different- 
colored gems. In the side space on the west, facing the east, 
they placed the different rush mats, with their variegated 
border; and the usual bench adorned with veined tortoise- 
shell. In the side space on the east, facing the west, they put 
the different mats of fine grass, with their border of painted 
silk; and the usual bench carved and adorned with gems. Be- 
fore the western side-chamber, facing the south, they placed 
the different mats of fine bamboo, with their dark mixed bor- 
der; and the usual lacquered bench. 

"They set forth also the five kinds of gems, and the pre- 
cious things of display. There were the red knife, the great 
lessons, the large convex symbol of gem, and the rounded and 
pointed maces .... 

"Two men in brownish leather caps, and holding three- 
cornered halberds, stood inside the gate leading to the private 
apartments: Four men in caps of spotted deer-skin, holding 
spears with upturned blades, stood one on each side of the 
steps east and west, and near to the platform of the hall. One 
man in a great officer's cap, and holding an axe, stood in the 
hall near the front at the east end. One man in a great officer's 
cap, and holding a somev/hat different axe, stood in the hall, 
near the front at the west end. One man in a great officer's 
cap, and holding a lance, stood at the front and eastern end 
of the hall. One man, in a great officer's cap, and holding a 
somewhat different lance, stood at the front and western end 
of the hall. One man in a great officer's cap, and holding a 
pointed weapon, stood by the steps on the north. " 

Here are mentioned many weapons, of different shapes, some 
similar to those in the Freer Gallery. In any case, we can confi- 
dently attribute our weapons to within a couple of generations 
of the Chou conquest, which would make their date looo B.C. 
plus or minus about 50 years, depending on which of the various 
chronological schemes one adopts. They very probably come from 
the Hsin-ts'un site, and possibly from the tomb of Marquis K'ang. 



68 



The ch'i broad axe (34.10), however, is clearly a hold-over 
from the Shang dynasty. Cheng Te-k'un (p. 242) points out that 
34.10 is the only example of this type of weapon to be recovered 
from an early Chou context, and that this is a characteristically 
Shang weapon type. In considering similar examples, Loehr ten- 
tatively dates them Shang and definitely dates number 6 in the 
Jannings catalogue as Shang (pp.20, 121, pl.V). These axes all 
have an asymmetrical tang, a fao-fieh on the tang and another 
on the blade, and a hole in the position of the mouth of the lower 
fao-fieh. So, our axe might be a Shang artifact or the translation 
of a Shang type into this early Chou context. 

The ko dagger axe 34.1 1, also bears affinities to Shang types, 
and while von Dewall classes it as a PV ko, the definite lack of a 
down-turning flange on the bottom of the blade seems to imply 
affinity to the PIII class, a class that belonged primarily to the 
Shang period but lingered into the Chou at Hsin-ts'un. It may be 
that the theory of a period of Shang-Chou transition when the 
Chou people were producing weapons and vessels modeled on 
Shang prototypes but with a definite Chou flavor is the correct 
explanation of the appearance of some of these weapons (Loehr, 
1968, p. 96; this idea was set forth in greater detail in an unpub- 
lished lecture by Virginia Kane "Bronze Vessels of the Shang- 
Chou Transition"). We must wait, however, for a complete tech- 
nical, stylistic, and epigraphical study of this group of weapons 
to be made before the whole story can be told. 

How does the use of iron in 34.10 and 34.1 1 relate to Chinese 
iron-casting technology? The easiest answer would be, "not at 
all." These two weapons are isolated instances of the use of iron 
as an ornamental device, just as jade was used in Shang times. The 
iron here appears as the blade, as does jade in the Freer ko 41.5 
and the axe or implement 41.4 (Freer Gallery of Art, 1946, 
pi. 43). This is entirely unlike the later use of iron for arrow shafts 
and agricultural implements. In fact, later in China, bronze was 
considered "the lovely metal" and iron "the ugly metal" (Need- 
ham, p. 2). While iron was also used for sumptuary arts, especially 
in belt-hooks, its major use was for ploughshares, tools, and so 



69 



forth. Kwang-chih Chang thinks that iron metallurgy in China 
probably dates from late in the period of the Spring and Autumn 
Annals (possibly around 600 B.C.) and the industry was in full 
swing in the early Warring states period (Chang, 1968, p. 313). 
This certainly corresponds with the archaeological evidence, and 
places our two weapons about 400 years before the wide industrial 
use of iron in China, clearly an anomalous position. 

Of course, as a number of writers make clear (see Needham, 
Chang Kwang-chih and Von der Merwe), the iron technology of 
early China is based on cast iron, and not wrought iron or steel. 
Needham states that "in the occidental world, the true iron age 
did not begin until about 1200 B.C., although terrestrial iron has 
been occasionally worked by man since a date of the order of 
2700 B.C." This places the beginning of the iron age in the Occi- 
dent at a time contemporary with the Shang dynasty in China. 
Iron appears in India around 800 B.C. (Banerjee, p. 239). So these 
weapons occur at a time before the introduction of iron foundries 
in China and India, and since there appears to have been no con- 
nection between the Near East and China, at least as far as the 
introduction of iron casting is concerned, these weapons are 
unique instances of the early use of iron in China, probably from 
a meteoritic source. These weapons must also have been unique in 
their own time, fit for a prince. 

There are many mentions of meteorite falls in ancient Chinese 
literature (Chang, Lapidarium Sinicum, p. 372-384). Most of 
these are simply stories of meteor sightings and their interpreta- 
tions as portents. If the metal in these weapons had been seen to 
fall from the sky it would certainly have been interpreted as aus- 
picious and this might be one reason for its use in the place of 
jade in these weapons. Such a use of meteoritic iron might also 
explain the fact that only one iron meteorite find is known from 
China. 

Of course, meteoritic iron, while rare, is by no means un- 
known to archaeology. The people of other cultures are known to 
have employed meteoritic iron to manufacture utensils and ob- 
jects of adornment even when they lived generally on a stone-age 



70 



level. The Eskimos of Northern Greenland fabricated harpoon- 
heads and scraping knives by hammering meteorite fragments 
and inserting them into bone handles (Buchwald and Munch, 
1965, p. II and figures 1-7). Several instances are known of fab- 
rication of plough-shares, machetes and nails of iron meteorites 
found by the native Mexican Indians and mestizos. The Hopewell 
Indians of North America are known to have made many objects 
of meteoritic iron, probably from the Brenham pallasite trans- 
ported from Kiowa County, Kansas (Wasson and Sedwick, 1969). 
The pioneers of 19th century North America often forged imple- 
ments of local meteoritic iron (Buchwald, 1965). It appears, in 
fact, that about 10 percent of all known iron meteorites have been 
wholly or partly heat-treated by man in attempts to split and 
utilize the material (V. F. Buchwald, personal communication). A 
detailed discussion of the history of many iron meteorites is in 
preparation by Buchwald. 

Many summaries of the use of meteorites by primitive man 
have appeared (Zimmer, 19 16; Richardson, 1934; Partington, 
1935; and Pokrzywnicki 1960/61). In these the emphasis is on 
weapons, but Richardson also mentions manufacture of amulets, 
beads, rings, and images. Smith (1965) points out that early man 
used new materials with their esthetic possibilities foremost in his 
mind. "For his delight man tries many things: for utility he em- 
ploys only what he knows will work." 

The final question that arises is the use of the casting-on tech- 
nique in the fabrication of these weapons. European metalworkers 
used this technique frequently in early periods (Drescher, 1 9 5 8, pi. 
I 5). In China, the technique was also used in bronze vessel produc- 
tion even before the period of these weapons. A tsun in the Freer 
collection (51.19) has animal-head decor which was apparently 
cast first and the vessel then cast on to it. A hu (49.5) has flanges 
which were inserted into the mold and the vessel was then cast on 
to them (Pope et aL, i^6y, pp. 102 and 43). This technique con- 
tinues into the Chou period, as can be seen on a yu (30.26, ihid., 
p. 355). It is interesting that the blades of these two weapons are 
cast on differently. That of 34.11 is chamfered like an arrow- 



71 



head, but that of 34.10 is drilled to provide recesses for inflow of 
molten bronze. 

Both methods apparently gave a sure attachment. The Chou 
artisan was accustomed to working with the casting-on technique 
and it must have seemed the natural method to use in this instance. 

A close examination of these two objects of the material cul- 
ture of ancient China has proved very instructive. We have been 
able to describe them in detail, to ascertain how they were made, 
and to relate them to their archaeological context. The proof of 
the use of meteoritic iron in these objects indicates the sensitivity 
of the ancient Chinese craftsman to materials, and in particular 
to the use of what must have been to him, a new and different 
material. Holding these aged, damaged and corroded objects in 
our hands, we appreciate the delight Marquis K'ang must have 
taken in owning them; and the establishment of this direct, tan- 
gible link to the mind of a man who lived three thousand years 
ago is one of the greatest pleasures an investigator can have. 



72 



BIBLIOGRAPHY 



Agrell, S. O., J. V. P. Long and R. E. Oglivie. "Nickel Content of Kamacite 
Near the Interface with Taenite in Iron Meteorites." Nature, vol. 198, 
1963, pp. 749-750- 

Axon, H.J. "Metallurgy of Meteorites." Progress in Materials Science, Bruce 
Chalmers and W. Hume Rothery, ed., vol. 13, no. 4, 1968, pp. 183-228. 

Banerjee, N. R. The Iron Age in India. Delhi, 1965. 

Birks, L. S. Electron Probe Microanalysis. New York 1963. 

Brill, Robert H. and Sheldon Moll. "The Electron-beam Probe Microana- 
lysis of Ancient Glass." Recent Advances in Conservation. G. Thomson, 
ed., London, 1963, pp. 145-1 51. 

Buchwald, Vagn F. "Heat Treated Iron Meteorites in Museum Collections." 
Geochimica et Cosmochimica Acta, vol. 29, 1965, pp. 603-604. 

— . "The Iron-nickel-phosphorus System and the Structure of Iron Mete- 
orites." Acta Polytechnica Scandinavica, 1966, ch. 51, pp. 1-46. 

— . "Study of Six Iron Meteorites." AnalectaGeologica, no. 2, 1967, pp. 1-75. 

Buchwald, V. F. and Sole Munch. "Catalogue of Meteorites." Analecta Geo- 
logica, vol. 1 , 1965, 81 pp. 

Buddhue, John Davis. The Oxidation and Weathering of Meteorites. Univer- 
sity of New Mexico publications in meteoritics no. 3, Albuquerque, 1957. 

Cassidy, William A. "The Wolf Creek, Western Australia, Meteorite Crater." 

Meteoritics, vol. i, 1954, pp. 197-199. 
Cassidy, William A., Luisa M. Villar, Theodore E. Bunch, Truman P. Koh- 

man and Daniel J. Milton. "Meteorites and Craters of Campo del Cielo, 

Argentina." Science, vol. 149, 1965, pp. 1055 — 1064. 
Castaing, R. "Application des sondes electroniques a une methode d'analyse 

ponctuelle chimique et cristallographique." Thesis, University of Paris, 

195 1. Also published in: Office National d'etudes et de recherches aero- 

nautique (ON ERA), no. 55, 1952, pp. 27-31. 
Chang, H. T. (Chang Hung Chao). Lapidarium Sinicum — A study of the 

rocks, fossils, and metals as known in Chinese literature (in Chinese). The 

Geological Society of China, Memoirs, Series B, no. 2. Peking, 1927. 
Chang Kwang-chih. TJje Archaeology of Ancient China. New Haven, 1968. 
Ch'en Meng-chia. ''Hsi-chou T'ung-ch'i Tuan-Tai.'' Part I. K'ao-ku Hshch- 

pao No. 9, (1955), pp. 161-165. Part VI. KKHPNo. 4 (1956), pi. 8. 



73 



— . "Malcolm's K'ang Hou Kuei and Its Set." Oriental Art, vol.1 (1948), 
pp. 1 1 i-i 16. 

Cheng Te-k'un. Chou China (Archaeology in China, Volume three). Cam- 
bridge, England, 1963. 
Ch'u Te-i. (Tch'ou To-yi) Bronzes antiques de la Chine. Paris 1924. 

Drescher, Hans. Der Uherfdngguss. Mainz, 1958. 

Feller-Kniepmeier, M. and H. H. Uhlig. "Nickel Analyses of Metallic Me- 
teorites by the Electron-probe Microanalycer." Geochimica et Cosmochi- 
micaActa, vol. 21, 1961, pp. 257-265. 

Freer Gallery of Art. A Descriptive and Illustrative Catalogue of Chinese 
Bronzes Acquired during the Administration of John Ellerton Lodge. 
Washington, 1946. 

Carrels, Robert M. and Charles L. Christ. Solutions, Minerals and Equi- 
libria. New York, 1965. 

Gettens, Rutherford John. Tloe Freer Chinese Bronzes, Volume II, Technical 
Studies. Freer Gallery of Art, Oriental Studies series. No. 7. Washington, 
1969. 

Goldstein, J. 1. and R. E. Ogilvie. "The Growth of the Widmanstatten 
Pattern in Metallic Meteorites." Geochimica et Cosmochimica Acta, vol. 
29, 1965, pp. 893-920. 

Hey, Max H. Catalogue of Meteorites. British Museum (Natural History), 
London, 1966. 

Hornblower, A. P. "Some Examples of Work with the Electron Probe 
Microanalyzer." Archaeometry, vol. 6, 1963, pp. 37-42. 

Keil, Klaus. "The Electron Microprobe X-ray Analyser and its Application 
in Mineralogy." Fortschritte der Mineralogie, vol. 44, 1967, pp. 4-66. 

Knox, Reed Jr. "Surviving Metal in Meteoritic Iron Oxide from the Wolf 
Creek, Western Australia, Meteorite Crater." Meteoritics, vol. 3, 1967, 
pp. 235-238. 

Kuo Pao-chiin. "Preliminary Report on the Excavations of the Ancient 
Cemetery at Hsin-ts'un, Hsiin Hsien, Honan" (in Chinese). K'ao-ku 
hsUeh-pao, 19^6, no. i, pp. 167-200. 

— . HsUn-hsien Hsin-ts'un (in Chinese). Peking, 1964. 

La Paz, Lincoln. "Meteorite Material from the Wolf Creek, Western Aus- 
tralia, Crater." Meteoritics, vol. i, 1954, pp. 200-203. 



74 



Lechtman, Heather. "Ancient Methods of Gilding Silver: Examples from the 
Old and the New Worlds." The Proceedings of the Fourth Symposium on 
Archaeological Chemistry, 1968. Robert Brill, ed., (in press). 

Legge, James. The Chinese Classics. London and Hongkong, not dated. 

Li Chi. "Typological Studies of the Bronze kou-ping (Chinese halberts) 
Excavated from Northern Honan, with a Classified and Illustrated List'" 
(in Chinese). Bulletin of TToe Institute of History and Philology, Acade- 
mia Sinica, vol. 22, 1959, pp. 1-18. 

Loehr, Max. Chinese Bronze Age Weapons. Ann Arbor, 1956. 

— . Ritual Vessels of Bronze Age China. New York, 1968. 

Mason, Brian. Meteorites. New York, 1962. 
Menzies, James Mellon. The Shang Ko. Toronto, 1965. 

Moseley, H. "The High-frequency Spectra of the Elements L" Philosophical 

Magazine, vol. 26, 191 3, pp. 1024-1034. 
Moss, A. A., M. H. Hey and D. L Bothwell. "Methods for the Chemical 

Analysis of Meteorites: L Siderites." Mineralogical Magazine, vol. 32, 

1961, pp. 802-816. 

Needham, Joseph. Tloe Development of Iron and Steel Technology in China. 
London, 1958. 

Nininger, H. H. Arizona's Meteorite Crater. Sedona, Arizona, 1956. 

Owen, E. A. and Y. H. Liu. "Further X-ray Study of the Equilibrium Dia- 
gram of the Iron-nickel System." Journal of the Iron and Steel Institute, 
vol. 163, 1949, pp. 132—137. 

Paneth, F. A. "The Discovery and Earliest Reproductions of the Widman- 
statten Figures." Geochimica et Cosmochimica Acta, vol. 18, i960, pp. 
176-182. 

Partington, J. R. Origins and Development of Applied Chemistry. London, 
1935- 

Pauly, Hans. "White Cast Iron with Cohenite, Schreibersite, and Sulphides 

from Tertiary Basalts on Disko, Greenland." Meddelelser fra Dansk 

Geologisk Forening, vol. 19, no. i, 1969, pp. 8-26. 
Perry, Stuart H. "The Metallography of Meteoritic Iron." Bulletin of the 

United States National Museum, vol. 184, 1944, pp. 1-204. 
Pokrzywnicki, J. "Weapons and other Articles made from Meteoritic Iron." 

Bulletin de la Societe des Amis des Sciences et des Lcttrcs de Poznan, 

series b, vol. XVI, 1960-61, pp. 25 1-277. 



75 



Pope, John Alexander, Rutherford John Gettens, James Cahill and Noel 
Barnard. The Freer Chinese Bronzes, Volume I, Catalogue. Freer Gallery 
of Art, Oriental Studies series. No. 7, Washington, 1967. 

Reed, S. J. B. "Electron-probe Microanalysis of the Metallic Phases in Iron 
yiQiQontQs.'' Geochimica et Cosmochimtca Acta, vol.29, 1965, pp. 535-549. 

Richardson, Harry Craig. "Iron Prehistoric and Ancient." American Journal 
of Achaeology, vol. 3 8 , 1 9 3 4, no. 4, pp. 5 5 5-5 8 3 . 

Short, James M. and Christian A. Andersen. "Electron Microprobe Ana- 
lyses of the Widmanstatten Structure in Nine Iron Meteorites." Journal 
of Geophysical Research, vol. 70, 1965, pp. 3745-3759. 

Smith, Cyril Stanley. A History of Metallography. Chicago, i960. 

— . "Note on the History of the Widmanstatten Structure." Geochimica et 
Cosmochimica Acta, vol. 26, 1962, pp. 971-972. 

— . "Materials and the Development of Civilization and Science." Science, 
vol. 148, 1965, pp. 908-917. 

Ta T'ang. "Chin-shih-chia Ch'u Te-i." In I-lin Ts'ung-lu, vol.2, p. 181, 
Hong Kong, 1962. 

Taylor, S. R. "The Wolf Creek Iron Meteorite." Nature, vol. 208, 1965, pp. 
944-945. 

van der Merwe, Johannes Nikolaas. The Metallurgical History and Carhon- 
14 Dating of Iron. Unpublished Ph. D. dissertation, Yale University, 
1967. 

von Dewall, Magdalene. "New Data on Early Chou Finds — their Relative 
Chronology in Historical Perspective." Symposium in Honor of Dr. Li 
Chi on his Seventieth Birthday — Part //. Taiwan, 1967, pp. 503-569. 

Waring, C. L. and C. S. Annell. "Semiquantitative Spectrographic Method 
for Analysis of Minerals, Rocks and Ores." Analytical Chemistry , vol. 
25, August 1953, pp. 1174-1179. 

Wasson, J. T. and S. P. Sedwick. "Possible Sources of Meteoritic Material 
from Hopewell Indian Burial Mounds." Nature, 111, 1^6^, pp. 22-24. 

White, John S., Jr., E. P. Henderson and Brian Mason. "Secondary Minerals 
Produced by Weathering of the Wolf Creek Meteorite." American Min- 
eralogist, yoL ^ 1, 196-/, p. 1190-1197. 

White, William Charles. Bronze Culture of Ancient China. Toronto, 1956. 

Wood, John A. "The Cooling Rates and Parent Planets of Several Iron 
Meteorites." Icarus, vol. 3, 1964, pp. 429-459. 



76 



Yetts, W. Perceval. "An Early Chou Bronze." The Burlington Magazine, 

vol. 70, April, 1937, no. 409, pp. 168-177. 
Young, William J. "Application of the Electron Microbeam Probe and Micro 

X-rays in Non-destructive Analysis." Recent advances in conservation. 

G.Thomson, ed., London, 1963, pp. 33-38. 

Zimmer, George Frederick. "The Use of Iron by Primitive Man." Journal 
of the Iron and Steel Institute, vol. 94, 1916, pp. 306-356. 



77 



SMITHSONIAN INSTITUTION LIBRARIES 




3 9088 01747 8363